checkpoint-11000
Browse files- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/README.md +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/adapter_config.json +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/adapter_model.safetensors +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/optimizer.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/rng_state.pth +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/scheduler.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/trainer_state.json +2208 -3
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/training_args.bin +0 -0
- qwen3.5-2b-50pct-masked-weighted/tb/events.out.tfevents.1790426135.b0aacb224392.4711.0 +2 -2
- qwen3.5-2b-50pct-masked-weighted/test_eval_entity/metrics.json +10 -10
- qwen3.5-2b-50pct-masked-weighted/test_eval_entity/predictions.jsonl +0 -0
- qwen3.5-2b-50pct-masked-weighted/test_eval_typo/metrics.json +11 -11
- qwen3.5-2b-50pct-masked-weighted/test_eval_typo/predictions.jsonl +0 -0
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/README.md
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/adapter_config.json
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/adapter_model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 134604152
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a67e62eeb53199d76d43705aca1752c800cea6a32563c5fbadef29b1e2fd57e5
|
| 3 |
size 134604152
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 269426431
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21cafaa1ec3bf9ace9c48e35428a55a94a76b2f8490a11267cd6138de0be4d78
|
| 3 |
size 269426431
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:238aeb030c3ac7c2245a02654e11e332dda22b33751b3ef6d1946bc1650516b5
|
| 3 |
size 14645
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e03ac1f05bcb978bdde6d20079e3e686a84c9d6f9c1eb919280bf02df27f1022
|
| 3 |
size 1465
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/trainer_state.json
RENAMED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 1.
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -5853,6 +5853,2211 @@
|
|
| 5853 |
"learning_rate": 2.972232716282584e-05,
|
| 5854 |
"loss": 0.5362387657165527,
|
| 5855 |
"step": 8000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5856 |
}
|
| 5857 |
],
|
| 5858 |
"logging_steps": 10,
|
|
@@ -5872,7 +8077,7 @@
|
|
| 5872 |
"attributes": {}
|
| 5873 |
}
|
| 5874 |
},
|
| 5875 |
-
"total_flos":
|
| 5876 |
"train_batch_size": 2,
|
| 5877 |
"trial_name": null,
|
| 5878 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.9739551786795881,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 11000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 5853 |
"learning_rate": 2.972232716282584e-05,
|
| 5854 |
"loss": 0.5362387657165527,
|
| 5855 |
"step": 8000
|
| 5856 |
+
},
|
| 5857 |
+
{
|
| 5858 |
+
"epoch": 1.4355609394979474,
|
| 5859 |
+
"step": 8000,
|
| 5860 |
+
"test_entity_exact_match": 0.0019801980198019802,
|
| 5861 |
+
"test_entity_fix_rate": 0.260669442965839,
|
| 5862 |
+
"test_entity_fix_rate_lenient": 0.2527750410509031,
|
| 5863 |
+
"test_entity_fix_rate_recoverable": 0.2565576235506399,
|
| 5864 |
+
"test_entity_hallucination_rate": 0.005940594059405941,
|
| 5865 |
+
"test_entity_preservation_rate": 0.7506771754865691,
|
| 5866 |
+
"test_entity_preservation_rate_lenient": 0.7525820150622891,
|
| 5867 |
+
"test_entity_targeted_score": 0.505673309226204,
|
| 5868 |
+
"test_entity_targeted_score_lenient": 0.5026785280565961,
|
| 5869 |
+
"test_entity_wer": 0.35609141661596266,
|
| 5870 |
+
"test_entity_wer_zwnj_normalized": 0.35184829367747994
|
| 5871 |
+
},
|
| 5872 |
+
{
|
| 5873 |
+
"epoch": 1.4355609394979474,
|
| 5874 |
+
"step": 8000,
|
| 5875 |
+
"test_typo_exact_match": 0.0111731843575419,
|
| 5876 |
+
"test_typo_fix_rate": 0.35459480418766964,
|
| 5877 |
+
"test_typo_fix_rate_lenient": 0.3473400952183813,
|
| 5878 |
+
"test_typo_fix_rate_recoverable": 0.3516183270281631,
|
| 5879 |
+
"test_typo_hallucination_rate": 0.013966480446927373,
|
| 5880 |
+
"test_typo_preservation_rate": 0.8494399200305766,
|
| 5881 |
+
"test_typo_preservation_rate_lenient": 0.8522077998183363,
|
| 5882 |
+
"test_typo_targeted_score": 0.6020173621091232,
|
| 5883 |
+
"test_typo_targeted_score_lenient": 0.5997739475183588,
|
| 5884 |
+
"test_typo_wer": 0.24783640958872635,
|
| 5885 |
+
"test_typo_wer_zwnj_normalized": 0.24222000970105434
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 1.4355609394979474,
|
| 5889 |
+
"step": 8000,
|
| 5890 |
+
"test_best_wer": 0.2231321104545041
|
| 5891 |
+
},
|
| 5892 |
+
{
|
| 5893 |
+
"epoch": 1.4373555869618861,
|
| 5894 |
+
"grad_norm": 1.8905603885650635,
|
| 5895 |
+
"learning_rate": 2.9627880619569327e-05,
|
| 5896 |
+
"loss": 0.6561440467834473,
|
| 5897 |
+
"step": 8010
|
| 5898 |
+
},
|
| 5899 |
+
{
|
| 5900 |
+
"epoch": 1.439150234425825,
|
| 5901 |
+
"grad_norm": 2.4211935997009277,
|
| 5902 |
+
"learning_rate": 2.953343407631281e-05,
|
| 5903 |
+
"loss": 0.7281019687652588,
|
| 5904 |
+
"step": 8020
|
| 5905 |
+
},
|
| 5906 |
+
{
|
| 5907 |
+
"epoch": 1.4409448818897639,
|
| 5908 |
+
"grad_norm": 0.8916682600975037,
|
| 5909 |
+
"learning_rate": 2.943898753305629e-05,
|
| 5910 |
+
"loss": 0.6491771697998047,
|
| 5911 |
+
"step": 8030
|
| 5912 |
+
},
|
| 5913 |
+
{
|
| 5914 |
+
"epoch": 1.4427395293537026,
|
| 5915 |
+
"grad_norm": 1.2187851667404175,
|
| 5916 |
+
"learning_rate": 2.9344540989799774e-05,
|
| 5917 |
+
"loss": 0.6241639137268067,
|
| 5918 |
+
"step": 8040
|
| 5919 |
+
},
|
| 5920 |
+
{
|
| 5921 |
+
"epoch": 1.4445341768176414,
|
| 5922 |
+
"grad_norm": 1.4738484621047974,
|
| 5923 |
+
"learning_rate": 2.9250094446543257e-05,
|
| 5924 |
+
"loss": 0.6121856689453125,
|
| 5925 |
+
"step": 8050
|
| 5926 |
+
},
|
| 5927 |
+
{
|
| 5928 |
+
"epoch": 1.4463288242815802,
|
| 5929 |
+
"grad_norm": 1.8711814880371094,
|
| 5930 |
+
"learning_rate": 2.915564790328674e-05,
|
| 5931 |
+
"loss": 0.6761098861694336,
|
| 5932 |
+
"step": 8060
|
| 5933 |
+
},
|
| 5934 |
+
{
|
| 5935 |
+
"epoch": 1.448123471745519,
|
| 5936 |
+
"grad_norm": 3.2736051082611084,
|
| 5937 |
+
"learning_rate": 2.9061201360030228e-05,
|
| 5938 |
+
"loss": 0.7006251335144043,
|
| 5939 |
+
"step": 8070
|
| 5940 |
+
},
|
| 5941 |
+
{
|
| 5942 |
+
"epoch": 1.4499181192094577,
|
| 5943 |
+
"grad_norm": 0.9859542846679688,
|
| 5944 |
+
"learning_rate": 2.8966754816773704e-05,
|
| 5945 |
+
"loss": 0.6910223484039306,
|
| 5946 |
+
"step": 8080
|
| 5947 |
+
},
|
| 5948 |
+
{
|
| 5949 |
+
"epoch": 1.4517127666733967,
|
| 5950 |
+
"grad_norm": 1.329181432723999,
|
| 5951 |
+
"learning_rate": 2.887230827351719e-05,
|
| 5952 |
+
"loss": 0.6295298099517822,
|
| 5953 |
+
"step": 8090
|
| 5954 |
+
},
|
| 5955 |
+
{
|
| 5956 |
+
"epoch": 1.4535074141373354,
|
| 5957 |
+
"grad_norm": 1.4134734869003296,
|
| 5958 |
+
"learning_rate": 2.8777861730260674e-05,
|
| 5959 |
+
"loss": 0.5724423408508301,
|
| 5960 |
+
"step": 8100
|
| 5961 |
+
},
|
| 5962 |
+
{
|
| 5963 |
+
"epoch": 1.4553020616012742,
|
| 5964 |
+
"grad_norm": 1.8962903022766113,
|
| 5965 |
+
"learning_rate": 2.8683415187004158e-05,
|
| 5966 |
+
"loss": 0.5908845901489258,
|
| 5967 |
+
"step": 8110
|
| 5968 |
+
},
|
| 5969 |
+
{
|
| 5970 |
+
"epoch": 1.457096709065213,
|
| 5971 |
+
"grad_norm": 3.4050114154815674,
|
| 5972 |
+
"learning_rate": 2.8588968643747638e-05,
|
| 5973 |
+
"loss": 0.8030138969421386,
|
| 5974 |
+
"step": 8120
|
| 5975 |
+
},
|
| 5976 |
+
{
|
| 5977 |
+
"epoch": 1.458891356529152,
|
| 5978 |
+
"grad_norm": 0.93116295337677,
|
| 5979 |
+
"learning_rate": 2.849452210049112e-05,
|
| 5980 |
+
"loss": 0.6308359146118164,
|
| 5981 |
+
"step": 8130
|
| 5982 |
+
},
|
| 5983 |
+
{
|
| 5984 |
+
"epoch": 1.4606860039930907,
|
| 5985 |
+
"grad_norm": 1.1376594305038452,
|
| 5986 |
+
"learning_rate": 2.8400075557234608e-05,
|
| 5987 |
+
"loss": 0.5804425239562988,
|
| 5988 |
+
"step": 8140
|
| 5989 |
+
},
|
| 5990 |
+
{
|
| 5991 |
+
"epoch": 1.4624806514570294,
|
| 5992 |
+
"grad_norm": 1.2171587944030762,
|
| 5993 |
+
"learning_rate": 2.830562901397809e-05,
|
| 5994 |
+
"loss": 0.5241493701934814,
|
| 5995 |
+
"step": 8150
|
| 5996 |
+
},
|
| 5997 |
+
{
|
| 5998 |
+
"epoch": 1.4642752989209682,
|
| 5999 |
+
"grad_norm": 1.9527809619903564,
|
| 6000 |
+
"learning_rate": 2.8211182470721575e-05,
|
| 6001 |
+
"loss": 0.6366949081420898,
|
| 6002 |
+
"step": 8160
|
| 6003 |
+
},
|
| 6004 |
+
{
|
| 6005 |
+
"epoch": 1.466069946384907,
|
| 6006 |
+
"grad_norm": 3.117816209793091,
|
| 6007 |
+
"learning_rate": 2.8116735927465055e-05,
|
| 6008 |
+
"loss": 0.7352989196777344,
|
| 6009 |
+
"step": 8170
|
| 6010 |
+
},
|
| 6011 |
+
{
|
| 6012 |
+
"epoch": 1.4678645938488457,
|
| 6013 |
+
"grad_norm": 1.0014796257019043,
|
| 6014 |
+
"learning_rate": 2.802228938420854e-05,
|
| 6015 |
+
"loss": 0.6335639953613281,
|
| 6016 |
+
"step": 8180
|
| 6017 |
+
},
|
| 6018 |
+
{
|
| 6019 |
+
"epoch": 1.4696592413127847,
|
| 6020 |
+
"grad_norm": 1.1855703592300415,
|
| 6021 |
+
"learning_rate": 2.7927842840952022e-05,
|
| 6022 |
+
"loss": 0.6076571941375732,
|
| 6023 |
+
"step": 8190
|
| 6024 |
+
},
|
| 6025 |
+
{
|
| 6026 |
+
"epoch": 1.4714538887767235,
|
| 6027 |
+
"grad_norm": 1.6133085489273071,
|
| 6028 |
+
"learning_rate": 2.783339629769551e-05,
|
| 6029 |
+
"loss": 0.6325051307678222,
|
| 6030 |
+
"step": 8200
|
| 6031 |
+
},
|
| 6032 |
+
{
|
| 6033 |
+
"epoch": 1.4732485362406622,
|
| 6034 |
+
"grad_norm": 2.0409483909606934,
|
| 6035 |
+
"learning_rate": 2.7738949754438985e-05,
|
| 6036 |
+
"loss": 0.6494714736938476,
|
| 6037 |
+
"step": 8210
|
| 6038 |
+
},
|
| 6039 |
+
{
|
| 6040 |
+
"epoch": 1.475043183704601,
|
| 6041 |
+
"grad_norm": 3.265942335128784,
|
| 6042 |
+
"learning_rate": 2.7644503211182472e-05,
|
| 6043 |
+
"loss": 0.6975038051605225,
|
| 6044 |
+
"step": 8220
|
| 6045 |
+
},
|
| 6046 |
+
{
|
| 6047 |
+
"epoch": 1.47683783116854,
|
| 6048 |
+
"grad_norm": 0.8910956978797913,
|
| 6049 |
+
"learning_rate": 2.7550056667925956e-05,
|
| 6050 |
+
"loss": 0.6698038101196289,
|
| 6051 |
+
"step": 8230
|
| 6052 |
+
},
|
| 6053 |
+
{
|
| 6054 |
+
"epoch": 1.4786324786324787,
|
| 6055 |
+
"grad_norm": 1.171595573425293,
|
| 6056 |
+
"learning_rate": 2.745561012466944e-05,
|
| 6057 |
+
"loss": 0.5993340969085693,
|
| 6058 |
+
"step": 8240
|
| 6059 |
+
},
|
| 6060 |
+
{
|
| 6061 |
+
"epoch": 1.4804271260964175,
|
| 6062 |
+
"grad_norm": 1.3274813890457153,
|
| 6063 |
+
"learning_rate": 2.7361163581412926e-05,
|
| 6064 |
+
"loss": 0.5612947463989257,
|
| 6065 |
+
"step": 8250
|
| 6066 |
+
},
|
| 6067 |
+
{
|
| 6068 |
+
"epoch": 1.4822217735603562,
|
| 6069 |
+
"grad_norm": 1.7055027484893799,
|
| 6070 |
+
"learning_rate": 2.7266717038156402e-05,
|
| 6071 |
+
"loss": 0.639337682723999,
|
| 6072 |
+
"step": 8260
|
| 6073 |
+
},
|
| 6074 |
+
{
|
| 6075 |
+
"epoch": 1.484016421024295,
|
| 6076 |
+
"grad_norm": 2.754523515701294,
|
| 6077 |
+
"learning_rate": 2.7172270494899886e-05,
|
| 6078 |
+
"loss": 0.692071008682251,
|
| 6079 |
+
"step": 8270
|
| 6080 |
+
},
|
| 6081 |
+
{
|
| 6082 |
+
"epoch": 1.4858110684882337,
|
| 6083 |
+
"grad_norm": 1.0503959655761719,
|
| 6084 |
+
"learning_rate": 2.7077823951643373e-05,
|
| 6085 |
+
"loss": 0.6727195739746094,
|
| 6086 |
+
"step": 8280
|
| 6087 |
+
},
|
| 6088 |
+
{
|
| 6089 |
+
"epoch": 1.4876057159521727,
|
| 6090 |
+
"grad_norm": 1.2824727296829224,
|
| 6091 |
+
"learning_rate": 2.6983377408386856e-05,
|
| 6092 |
+
"loss": 0.5498480319976806,
|
| 6093 |
+
"step": 8290
|
| 6094 |
+
},
|
| 6095 |
+
{
|
| 6096 |
+
"epoch": 1.4894003634161115,
|
| 6097 |
+
"grad_norm": 1.6015315055847168,
|
| 6098 |
+
"learning_rate": 2.6888930865130336e-05,
|
| 6099 |
+
"loss": 0.5700726509094238,
|
| 6100 |
+
"step": 8300
|
| 6101 |
+
},
|
| 6102 |
+
{
|
| 6103 |
+
"epoch": 1.4911950108800502,
|
| 6104 |
+
"grad_norm": 1.7487208843231201,
|
| 6105 |
+
"learning_rate": 2.679448432187382e-05,
|
| 6106 |
+
"loss": 0.6561488151550293,
|
| 6107 |
+
"step": 8310
|
| 6108 |
+
},
|
| 6109 |
+
{
|
| 6110 |
+
"epoch": 1.492989658343989,
|
| 6111 |
+
"grad_norm": 2.881220579147339,
|
| 6112 |
+
"learning_rate": 2.6700037778617303e-05,
|
| 6113 |
+
"loss": 0.6268972873687744,
|
| 6114 |
+
"step": 8320
|
| 6115 |
+
},
|
| 6116 |
+
{
|
| 6117 |
+
"epoch": 1.494784305807928,
|
| 6118 |
+
"grad_norm": 1.033646821975708,
|
| 6119 |
+
"learning_rate": 2.660559123536079e-05,
|
| 6120 |
+
"loss": 0.6984382152557373,
|
| 6121 |
+
"step": 8330
|
| 6122 |
+
},
|
| 6123 |
+
{
|
| 6124 |
+
"epoch": 1.4965789532718667,
|
| 6125 |
+
"grad_norm": 1.104401707649231,
|
| 6126 |
+
"learning_rate": 2.6511144692104273e-05,
|
| 6127 |
+
"loss": 0.556285047531128,
|
| 6128 |
+
"step": 8340
|
| 6129 |
+
},
|
| 6130 |
+
{
|
| 6131 |
+
"epoch": 1.4983736007358055,
|
| 6132 |
+
"grad_norm": 1.291401743888855,
|
| 6133 |
+
"learning_rate": 2.6416698148847753e-05,
|
| 6134 |
+
"loss": 0.6780119419097901,
|
| 6135 |
+
"step": 8350
|
| 6136 |
+
},
|
| 6137 |
+
{
|
| 6138 |
+
"epoch": 1.5001682481997443,
|
| 6139 |
+
"grad_norm": 1.695534110069275,
|
| 6140 |
+
"learning_rate": 2.6322251605591237e-05,
|
| 6141 |
+
"loss": 0.6268007278442382,
|
| 6142 |
+
"step": 8360
|
| 6143 |
+
},
|
| 6144 |
+
{
|
| 6145 |
+
"epoch": 1.501962895663683,
|
| 6146 |
+
"grad_norm": 2.7133352756500244,
|
| 6147 |
+
"learning_rate": 2.622780506233472e-05,
|
| 6148 |
+
"loss": 0.6975327014923096,
|
| 6149 |
+
"step": 8370
|
| 6150 |
+
},
|
| 6151 |
+
{
|
| 6152 |
+
"epoch": 1.5037575431276218,
|
| 6153 |
+
"grad_norm": 0.9616020321846008,
|
| 6154 |
+
"learning_rate": 2.6133358519078204e-05,
|
| 6155 |
+
"loss": 0.652420949935913,
|
| 6156 |
+
"step": 8380
|
| 6157 |
+
},
|
| 6158 |
+
{
|
| 6159 |
+
"epoch": 1.5055521905915605,
|
| 6160 |
+
"grad_norm": 1.309381127357483,
|
| 6161 |
+
"learning_rate": 2.6038911975821684e-05,
|
| 6162 |
+
"loss": 0.6299611568450928,
|
| 6163 |
+
"step": 8390
|
| 6164 |
+
},
|
| 6165 |
+
{
|
| 6166 |
+
"epoch": 1.5073468380554995,
|
| 6167 |
+
"grad_norm": 1.4106836318969727,
|
| 6168 |
+
"learning_rate": 2.5944465432565167e-05,
|
| 6169 |
+
"loss": 0.6366672039031982,
|
| 6170 |
+
"step": 8400
|
| 6171 |
+
},
|
| 6172 |
+
{
|
| 6173 |
+
"epoch": 1.5091414855194383,
|
| 6174 |
+
"grad_norm": 1.63096284866333,
|
| 6175 |
+
"learning_rate": 2.5850018889308654e-05,
|
| 6176 |
+
"loss": 0.6294029235839844,
|
| 6177 |
+
"step": 8410
|
| 6178 |
+
},
|
| 6179 |
+
{
|
| 6180 |
+
"epoch": 1.510936132983377,
|
| 6181 |
+
"grad_norm": 3.0822575092315674,
|
| 6182 |
+
"learning_rate": 2.5755572346052137e-05,
|
| 6183 |
+
"loss": 0.6654938697814942,
|
| 6184 |
+
"step": 8420
|
| 6185 |
+
},
|
| 6186 |
+
{
|
| 6187 |
+
"epoch": 1.512730780447316,
|
| 6188 |
+
"grad_norm": 1.0821377038955688,
|
| 6189 |
+
"learning_rate": 2.566112580279562e-05,
|
| 6190 |
+
"loss": 0.6763864040374756,
|
| 6191 |
+
"step": 8430
|
| 6192 |
+
},
|
| 6193 |
+
{
|
| 6194 |
+
"epoch": 1.5145254279112548,
|
| 6195 |
+
"grad_norm": 1.1002248525619507,
|
| 6196 |
+
"learning_rate": 2.55666792595391e-05,
|
| 6197 |
+
"loss": 0.5553943634033203,
|
| 6198 |
+
"step": 8440
|
| 6199 |
+
},
|
| 6200 |
+
{
|
| 6201 |
+
"epoch": 1.5163200753751935,
|
| 6202 |
+
"grad_norm": 1.2352007627487183,
|
| 6203 |
+
"learning_rate": 2.5472232716282584e-05,
|
| 6204 |
+
"loss": 0.5612579822540283,
|
| 6205 |
+
"step": 8450
|
| 6206 |
+
},
|
| 6207 |
+
{
|
| 6208 |
+
"epoch": 1.5181147228391323,
|
| 6209 |
+
"grad_norm": 1.8368786573410034,
|
| 6210 |
+
"learning_rate": 2.5377786173026068e-05,
|
| 6211 |
+
"loss": 0.662775182723999,
|
| 6212 |
+
"step": 8460
|
| 6213 |
+
},
|
| 6214 |
+
{
|
| 6215 |
+
"epoch": 1.519909370303071,
|
| 6216 |
+
"grad_norm": 2.6937034130096436,
|
| 6217 |
+
"learning_rate": 2.5283339629769554e-05,
|
| 6218 |
+
"loss": 0.6988917350769043,
|
| 6219 |
+
"step": 8470
|
| 6220 |
+
},
|
| 6221 |
+
{
|
| 6222 |
+
"epoch": 1.5217040177670098,
|
| 6223 |
+
"grad_norm": 0.9936008453369141,
|
| 6224 |
+
"learning_rate": 2.518889308651303e-05,
|
| 6225 |
+
"loss": 0.6265581607818603,
|
| 6226 |
+
"step": 8480
|
| 6227 |
+
},
|
| 6228 |
+
{
|
| 6229 |
+
"epoch": 1.5234986652309486,
|
| 6230 |
+
"grad_norm": 1.2112939357757568,
|
| 6231 |
+
"learning_rate": 2.5094446543256518e-05,
|
| 6232 |
+
"loss": 0.5488441944122314,
|
| 6233 |
+
"step": 8490
|
| 6234 |
+
},
|
| 6235 |
+
{
|
| 6236 |
+
"epoch": 1.5252933126948875,
|
| 6237 |
+
"grad_norm": 1.2531001567840576,
|
| 6238 |
+
"learning_rate": 2.5e-05,
|
| 6239 |
+
"loss": 0.5412239551544189,
|
| 6240 |
+
"step": 8500
|
| 6241 |
+
},
|
| 6242 |
+
{
|
| 6243 |
+
"epoch": 1.5270879601588263,
|
| 6244 |
+
"grad_norm": 1.7842986583709717,
|
| 6245 |
+
"learning_rate": 2.4905553456743485e-05,
|
| 6246 |
+
"loss": 0.6409042835235595,
|
| 6247 |
+
"step": 8510
|
| 6248 |
+
},
|
| 6249 |
+
{
|
| 6250 |
+
"epoch": 1.528882607622765,
|
| 6251 |
+
"grad_norm": 3.35718035697937,
|
| 6252 |
+
"learning_rate": 2.4811106913486968e-05,
|
| 6253 |
+
"loss": 0.6992044448852539,
|
| 6254 |
+
"step": 8520
|
| 6255 |
+
},
|
| 6256 |
+
{
|
| 6257 |
+
"epoch": 1.530677255086704,
|
| 6258 |
+
"grad_norm": 1.069508671760559,
|
| 6259 |
+
"learning_rate": 2.471666037023045e-05,
|
| 6260 |
+
"loss": 0.6775628566741944,
|
| 6261 |
+
"step": 8530
|
| 6262 |
+
},
|
| 6263 |
+
{
|
| 6264 |
+
"epoch": 1.5324719025506428,
|
| 6265 |
+
"grad_norm": 1.230456829071045,
|
| 6266 |
+
"learning_rate": 2.4622213826973935e-05,
|
| 6267 |
+
"loss": 0.5915063858032227,
|
| 6268 |
+
"step": 8540
|
| 6269 |
+
},
|
| 6270 |
+
{
|
| 6271 |
+
"epoch": 1.5342665500145816,
|
| 6272 |
+
"grad_norm": 1.4768152236938477,
|
| 6273 |
+
"learning_rate": 2.452776728371742e-05,
|
| 6274 |
+
"loss": 0.5757005214691162,
|
| 6275 |
+
"step": 8550
|
| 6276 |
+
},
|
| 6277 |
+
{
|
| 6278 |
+
"epoch": 1.5360611974785203,
|
| 6279 |
+
"grad_norm": 1.7781312465667725,
|
| 6280 |
+
"learning_rate": 2.44333207404609e-05,
|
| 6281 |
+
"loss": 0.6322662830352783,
|
| 6282 |
+
"step": 8560
|
| 6283 |
+
},
|
| 6284 |
+
{
|
| 6285 |
+
"epoch": 1.537855844942459,
|
| 6286 |
+
"grad_norm": 3.024894952774048,
|
| 6287 |
+
"learning_rate": 2.4338874197204385e-05,
|
| 6288 |
+
"loss": 0.6350002765655518,
|
| 6289 |
+
"step": 8570
|
| 6290 |
+
},
|
| 6291 |
+
{
|
| 6292 |
+
"epoch": 1.5396504924063978,
|
| 6293 |
+
"grad_norm": 0.9935160279273987,
|
| 6294 |
+
"learning_rate": 2.4244427653947865e-05,
|
| 6295 |
+
"loss": 0.7224649429321289,
|
| 6296 |
+
"step": 8580
|
| 6297 |
+
},
|
| 6298 |
+
{
|
| 6299 |
+
"epoch": 1.5414451398703366,
|
| 6300 |
+
"grad_norm": 1.122530460357666,
|
| 6301 |
+
"learning_rate": 2.414998111069135e-05,
|
| 6302 |
+
"loss": 0.5923322200775146,
|
| 6303 |
+
"step": 8590
|
| 6304 |
+
},
|
| 6305 |
+
{
|
| 6306 |
+
"epoch": 1.5432397873342754,
|
| 6307 |
+
"grad_norm": 1.3914778232574463,
|
| 6308 |
+
"learning_rate": 2.4055534567434835e-05,
|
| 6309 |
+
"loss": 0.5903297901153565,
|
| 6310 |
+
"step": 8600
|
| 6311 |
+
},
|
| 6312 |
+
{
|
| 6313 |
+
"epoch": 1.5450344347982143,
|
| 6314 |
+
"grad_norm": 1.5583442449569702,
|
| 6315 |
+
"learning_rate": 2.3961088024178315e-05,
|
| 6316 |
+
"loss": 0.5861922264099121,
|
| 6317 |
+
"step": 8610
|
| 6318 |
+
},
|
| 6319 |
+
{
|
| 6320 |
+
"epoch": 1.546829082262153,
|
| 6321 |
+
"grad_norm": 3.0478203296661377,
|
| 6322 |
+
"learning_rate": 2.38666414809218e-05,
|
| 6323 |
+
"loss": 0.7032251834869385,
|
| 6324 |
+
"step": 8620
|
| 6325 |
+
},
|
| 6326 |
+
{
|
| 6327 |
+
"epoch": 1.548623729726092,
|
| 6328 |
+
"grad_norm": 0.9380120038986206,
|
| 6329 |
+
"learning_rate": 2.3772194937665282e-05,
|
| 6330 |
+
"loss": 0.693181848526001,
|
| 6331 |
+
"step": 8630
|
| 6332 |
+
},
|
| 6333 |
+
{
|
| 6334 |
+
"epoch": 1.5504183771900308,
|
| 6335 |
+
"grad_norm": 1.2389183044433594,
|
| 6336 |
+
"learning_rate": 2.3677748394408766e-05,
|
| 6337 |
+
"loss": 0.6143226146697998,
|
| 6338 |
+
"step": 8640
|
| 6339 |
+
},
|
| 6340 |
+
{
|
| 6341 |
+
"epoch": 1.5522130246539696,
|
| 6342 |
+
"grad_norm": 1.522462248802185,
|
| 6343 |
+
"learning_rate": 2.358330185115225e-05,
|
| 6344 |
+
"loss": 0.5398208141326905,
|
| 6345 |
+
"step": 8650
|
| 6346 |
+
},
|
| 6347 |
+
{
|
| 6348 |
+
"epoch": 1.5540076721179084,
|
| 6349 |
+
"grad_norm": 1.609066367149353,
|
| 6350 |
+
"learning_rate": 2.3488855307895733e-05,
|
| 6351 |
+
"loss": 0.6074576377868652,
|
| 6352 |
+
"step": 8660
|
| 6353 |
+
},
|
| 6354 |
+
{
|
| 6355 |
+
"epoch": 1.5558023195818471,
|
| 6356 |
+
"grad_norm": 3.6447341442108154,
|
| 6357 |
+
"learning_rate": 2.3394408764639213e-05,
|
| 6358 |
+
"loss": 0.7698046684265136,
|
| 6359 |
+
"step": 8670
|
| 6360 |
+
},
|
| 6361 |
+
{
|
| 6362 |
+
"epoch": 1.5575969670457859,
|
| 6363 |
+
"grad_norm": 0.8544888496398926,
|
| 6364 |
+
"learning_rate": 2.32999622213827e-05,
|
| 6365 |
+
"loss": 0.6298615455627441,
|
| 6366 |
+
"step": 8680
|
| 6367 |
+
},
|
| 6368 |
+
{
|
| 6369 |
+
"epoch": 1.5593916145097246,
|
| 6370 |
+
"grad_norm": 1.1165138483047485,
|
| 6371 |
+
"learning_rate": 2.3205515678126183e-05,
|
| 6372 |
+
"loss": 0.5594109058380127,
|
| 6373 |
+
"step": 8690
|
| 6374 |
+
},
|
| 6375 |
+
{
|
| 6376 |
+
"epoch": 1.5611862619736634,
|
| 6377 |
+
"grad_norm": 1.4863481521606445,
|
| 6378 |
+
"learning_rate": 2.3111069134869663e-05,
|
| 6379 |
+
"loss": 0.5747323989868164,
|
| 6380 |
+
"step": 8700
|
| 6381 |
+
},
|
| 6382 |
+
{
|
| 6383 |
+
"epoch": 1.5629809094376024,
|
| 6384 |
+
"grad_norm": 1.5945227146148682,
|
| 6385 |
+
"learning_rate": 2.301662259161315e-05,
|
| 6386 |
+
"loss": 0.6580572128295898,
|
| 6387 |
+
"step": 8710
|
| 6388 |
+
},
|
| 6389 |
+
{
|
| 6390 |
+
"epoch": 1.5647755569015411,
|
| 6391 |
+
"grad_norm": 3.058377981185913,
|
| 6392 |
+
"learning_rate": 2.292217604835663e-05,
|
| 6393 |
+
"loss": 0.6520285606384277,
|
| 6394 |
+
"step": 8720
|
| 6395 |
+
},
|
| 6396 |
+
{
|
| 6397 |
+
"epoch": 1.5665702043654801,
|
| 6398 |
+
"grad_norm": 0.9069721102714539,
|
| 6399 |
+
"learning_rate": 2.2827729505100117e-05,
|
| 6400 |
+
"loss": 0.6168886661529541,
|
| 6401 |
+
"step": 8730
|
| 6402 |
+
},
|
| 6403 |
+
{
|
| 6404 |
+
"epoch": 1.5683648518294189,
|
| 6405 |
+
"grad_norm": 1.0873689651489258,
|
| 6406 |
+
"learning_rate": 2.2733282961843597e-05,
|
| 6407 |
+
"loss": 0.5469409465789795,
|
| 6408 |
+
"step": 8740
|
| 6409 |
+
},
|
| 6410 |
+
{
|
| 6411 |
+
"epoch": 1.5701594992933576,
|
| 6412 |
+
"grad_norm": 1.2975069284439087,
|
| 6413 |
+
"learning_rate": 2.263883641858708e-05,
|
| 6414 |
+
"loss": 0.5552399635314942,
|
| 6415 |
+
"step": 8750
|
| 6416 |
+
},
|
| 6417 |
+
{
|
| 6418 |
+
"epoch": 1.5719541467572964,
|
| 6419 |
+
"grad_norm": 1.9046076536178589,
|
| 6420 |
+
"learning_rate": 2.2544389875330563e-05,
|
| 6421 |
+
"loss": 0.6128366947174072,
|
| 6422 |
+
"step": 8760
|
| 6423 |
+
},
|
| 6424 |
+
{
|
| 6425 |
+
"epoch": 1.5737487942212351,
|
| 6426 |
+
"grad_norm": 3.9565863609313965,
|
| 6427 |
+
"learning_rate": 2.2449943332074047e-05,
|
| 6428 |
+
"loss": 0.7613388538360596,
|
| 6429 |
+
"step": 8770
|
| 6430 |
+
},
|
| 6431 |
+
{
|
| 6432 |
+
"epoch": 1.575543441685174,
|
| 6433 |
+
"grad_norm": 0.9836629629135132,
|
| 6434 |
+
"learning_rate": 2.235549678881753e-05,
|
| 6435 |
+
"loss": 0.609618854522705,
|
| 6436 |
+
"step": 8780
|
| 6437 |
+
},
|
| 6438 |
+
{
|
| 6439 |
+
"epoch": 1.5773380891491127,
|
| 6440 |
+
"grad_norm": 1.1420564651489258,
|
| 6441 |
+
"learning_rate": 2.2261050245561014e-05,
|
| 6442 |
+
"loss": 0.562946081161499,
|
| 6443 |
+
"step": 8790
|
| 6444 |
+
},
|
| 6445 |
+
{
|
| 6446 |
+
"epoch": 1.5791327366130514,
|
| 6447 |
+
"grad_norm": 1.4746840000152588,
|
| 6448 |
+
"learning_rate": 2.2166603702304497e-05,
|
| 6449 |
+
"loss": 0.5291254043579101,
|
| 6450 |
+
"step": 8800
|
| 6451 |
+
},
|
| 6452 |
+
{
|
| 6453 |
+
"epoch": 1.5809273840769904,
|
| 6454 |
+
"grad_norm": 1.7385553121566772,
|
| 6455 |
+
"learning_rate": 2.207215715904798e-05,
|
| 6456 |
+
"loss": 0.6076802730560302,
|
| 6457 |
+
"step": 8810
|
| 6458 |
+
},
|
| 6459 |
+
{
|
| 6460 |
+
"epoch": 1.5827220315409292,
|
| 6461 |
+
"grad_norm": 2.885716199874878,
|
| 6462 |
+
"learning_rate": 2.1977710615791464e-05,
|
| 6463 |
+
"loss": 0.6692715167999268,
|
| 6464 |
+
"step": 8820
|
| 6465 |
+
},
|
| 6466 |
+
{
|
| 6467 |
+
"epoch": 1.5845166790048681,
|
| 6468 |
+
"grad_norm": 0.8791989088058472,
|
| 6469 |
+
"learning_rate": 2.1883264072534944e-05,
|
| 6470 |
+
"loss": 0.6661964893341065,
|
| 6471 |
+
"step": 8830
|
| 6472 |
+
},
|
| 6473 |
+
{
|
| 6474 |
+
"epoch": 1.586311326468807,
|
| 6475 |
+
"grad_norm": 1.2418320178985596,
|
| 6476 |
+
"learning_rate": 2.178881752927843e-05,
|
| 6477 |
+
"loss": 0.5952839851379395,
|
| 6478 |
+
"step": 8840
|
| 6479 |
+
},
|
| 6480 |
+
{
|
| 6481 |
+
"epoch": 1.5881059739327457,
|
| 6482 |
+
"grad_norm": 1.3459867238998413,
|
| 6483 |
+
"learning_rate": 2.169437098602191e-05,
|
| 6484 |
+
"loss": 0.5728863716125489,
|
| 6485 |
+
"step": 8850
|
| 6486 |
+
},
|
| 6487 |
+
{
|
| 6488 |
+
"epoch": 1.5899006213966844,
|
| 6489 |
+
"grad_norm": 2.1040303707122803,
|
| 6490 |
+
"learning_rate": 2.1599924442765394e-05,
|
| 6491 |
+
"loss": 0.5795969486236572,
|
| 6492 |
+
"step": 8860
|
| 6493 |
+
},
|
| 6494 |
+
{
|
| 6495 |
+
"epoch": 1.5916952688606232,
|
| 6496 |
+
"grad_norm": 2.6528022289276123,
|
| 6497 |
+
"learning_rate": 2.150547789950888e-05,
|
| 6498 |
+
"loss": 0.742084550857544,
|
| 6499 |
+
"step": 8870
|
| 6500 |
+
},
|
| 6501 |
+
{
|
| 6502 |
+
"epoch": 1.593489916324562,
|
| 6503 |
+
"grad_norm": 1.0152089595794678,
|
| 6504 |
+
"learning_rate": 2.141103135625236e-05,
|
| 6505 |
+
"loss": 0.6684091567993165,
|
| 6506 |
+
"step": 8880
|
| 6507 |
+
},
|
| 6508 |
+
{
|
| 6509 |
+
"epoch": 1.5952845637885007,
|
| 6510 |
+
"grad_norm": 1.1930161714553833,
|
| 6511 |
+
"learning_rate": 2.1316584812995848e-05,
|
| 6512 |
+
"loss": 0.5855762481689453,
|
| 6513 |
+
"step": 8890
|
| 6514 |
+
},
|
| 6515 |
+
{
|
| 6516 |
+
"epoch": 1.5970792112524395,
|
| 6517 |
+
"grad_norm": 1.371364712715149,
|
| 6518 |
+
"learning_rate": 2.1222138269739328e-05,
|
| 6519 |
+
"loss": 0.5302097797393799,
|
| 6520 |
+
"step": 8900
|
| 6521 |
+
},
|
| 6522 |
+
{
|
| 6523 |
+
"epoch": 1.5988738587163784,
|
| 6524 |
+
"grad_norm": 1.6537835597991943,
|
| 6525 |
+
"learning_rate": 2.112769172648281e-05,
|
| 6526 |
+
"loss": 0.5780539035797119,
|
| 6527 |
+
"step": 8910
|
| 6528 |
+
},
|
| 6529 |
+
{
|
| 6530 |
+
"epoch": 1.6006685061803172,
|
| 6531 |
+
"grad_norm": 2.7254743576049805,
|
| 6532 |
+
"learning_rate": 2.1033245183226295e-05,
|
| 6533 |
+
"loss": 0.6588037967681885,
|
| 6534 |
+
"step": 8920
|
| 6535 |
+
},
|
| 6536 |
+
{
|
| 6537 |
+
"epoch": 1.602463153644256,
|
| 6538 |
+
"grad_norm": 1.0531034469604492,
|
| 6539 |
+
"learning_rate": 2.0938798639969778e-05,
|
| 6540 |
+
"loss": 0.7268345832824707,
|
| 6541 |
+
"step": 8930
|
| 6542 |
+
},
|
| 6543 |
+
{
|
| 6544 |
+
"epoch": 1.604257801108195,
|
| 6545 |
+
"grad_norm": 1.15487802028656,
|
| 6546 |
+
"learning_rate": 2.084435209671326e-05,
|
| 6547 |
+
"loss": 0.5964405059814453,
|
| 6548 |
+
"step": 8940
|
| 6549 |
+
},
|
| 6550 |
+
{
|
| 6551 |
+
"epoch": 1.6060524485721337,
|
| 6552 |
+
"grad_norm": 1.3200647830963135,
|
| 6553 |
+
"learning_rate": 2.0749905553456745e-05,
|
| 6554 |
+
"loss": 0.5342237472534179,
|
| 6555 |
+
"step": 8950
|
| 6556 |
+
},
|
| 6557 |
+
{
|
| 6558 |
+
"epoch": 1.6078470960360725,
|
| 6559 |
+
"grad_norm": 1.7189518213272095,
|
| 6560 |
+
"learning_rate": 2.065545901020023e-05,
|
| 6561 |
+
"loss": 0.5573340892791748,
|
| 6562 |
+
"step": 8960
|
| 6563 |
+
},
|
| 6564 |
+
{
|
| 6565 |
+
"epoch": 1.6096417435000112,
|
| 6566 |
+
"grad_norm": 2.760460138320923,
|
| 6567 |
+
"learning_rate": 2.0561012466943712e-05,
|
| 6568 |
+
"loss": 0.6734081745147705,
|
| 6569 |
+
"step": 8970
|
| 6570 |
+
},
|
| 6571 |
+
{
|
| 6572 |
+
"epoch": 1.61143639096395,
|
| 6573 |
+
"grad_norm": 0.9920043349266052,
|
| 6574 |
+
"learning_rate": 2.0466565923687195e-05,
|
| 6575 |
+
"loss": 0.6704621315002441,
|
| 6576 |
+
"step": 8980
|
| 6577 |
+
},
|
| 6578 |
+
{
|
| 6579 |
+
"epoch": 1.6132310384278887,
|
| 6580 |
+
"grad_norm": 1.1208456754684448,
|
| 6581 |
+
"learning_rate": 2.0372119380430675e-05,
|
| 6582 |
+
"loss": 0.5830750465393066,
|
| 6583 |
+
"step": 8990
|
| 6584 |
+
},
|
| 6585 |
+
{
|
| 6586 |
+
"epoch": 1.6150256858918275,
|
| 6587 |
+
"grad_norm": 1.4231200218200684,
|
| 6588 |
+
"learning_rate": 2.0277672837174162e-05,
|
| 6589 |
+
"loss": 0.5520272731781006,
|
| 6590 |
+
"step": 9000
|
| 6591 |
+
},
|
| 6592 |
+
{
|
| 6593 |
+
"epoch": 1.6150256858918275,
|
| 6594 |
+
"step": 9000,
|
| 6595 |
+
"test_entity_exact_match": 0.0029702970297029703,
|
| 6596 |
+
"test_entity_fix_rate": 0.2628925697466888,
|
| 6597 |
+
"test_entity_fix_rate_lenient": 0.2548768472906404,
|
| 6598 |
+
"test_entity_fix_rate_recoverable": 0.25872957529989643,
|
| 6599 |
+
"test_entity_hallucination_rate": 0.01485148514851485,
|
| 6600 |
+
"test_entity_preservation_rate": 0.7392182724786874,
|
| 6601 |
+
"test_entity_preservation_rate_lenient": 0.7415418960009238,
|
| 6602 |
+
"test_entity_targeted_score": 0.5010554211126881,
|
| 6603 |
+
"test_entity_targeted_score_lenient": 0.4982093716457821,
|
| 6604 |
+
"test_entity_wer": 0.3787267428827617,
|
| 6605 |
+
"test_entity_wer_zwnj_normalized": 0.3748038556377494
|
| 6606 |
+
},
|
| 6607 |
+
{
|
| 6608 |
+
"epoch": 1.6150256858918275,
|
| 6609 |
+
"step": 9000,
|
| 6610 |
+
"test_typo_exact_match": 0.013966480446927373,
|
| 6611 |
+
"test_typo_fix_rate": 0.35847227607599846,
|
| 6612 |
+
"test_typo_fix_rate_lenient": 0.35003104947215896,
|
| 6613 |
+
"test_typo_fix_rate_recoverable": 0.35435056746532156,
|
| 6614 |
+
"test_typo_hallucination_rate": 0.0223463687150838,
|
| 6615 |
+
"test_typo_preservation_rate": 0.8486755064240143,
|
| 6616 |
+
"test_typo_preservation_rate_lenient": 0.8516803891118989,
|
| 6617 |
+
"test_typo_targeted_score": 0.6035738912500064,
|
| 6618 |
+
"test_typo_targeted_score_lenient": 0.600855719292029,
|
| 6619 |
+
"test_typo_wer": 0.26823415281713514,
|
| 6620 |
+
"test_typo_wer_zwnj_normalized": 0.26307727655663626
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 1.6150256858918275,
|
| 6624 |
+
"step": 9000,
|
| 6625 |
+
"test_best_wer": 0.2231321104545041
|
| 6626 |
+
},
|
| 6627 |
+
{
|
| 6628 |
+
"epoch": 1.6168203333557665,
|
| 6629 |
+
"grad_norm": 1.8826247453689575,
|
| 6630 |
+
"learning_rate": 2.0183226293917642e-05,
|
| 6631 |
+
"loss": 0.6087816238403321,
|
| 6632 |
+
"step": 9010
|
| 6633 |
+
},
|
| 6634 |
+
{
|
| 6635 |
+
"epoch": 1.6186149808197052,
|
| 6636 |
+
"grad_norm": 2.611398458480835,
|
| 6637 |
+
"learning_rate": 2.0088779750661126e-05,
|
| 6638 |
+
"loss": 0.6654233455657959,
|
| 6639 |
+
"step": 9020
|
| 6640 |
+
},
|
| 6641 |
+
{
|
| 6642 |
+
"epoch": 1.620409628283644,
|
| 6643 |
+
"grad_norm": 1.040740728378296,
|
| 6644 |
+
"learning_rate": 1.999433320740461e-05,
|
| 6645 |
+
"loss": 0.6901034355163574,
|
| 6646 |
+
"step": 9030
|
| 6647 |
+
},
|
| 6648 |
+
{
|
| 6649 |
+
"epoch": 1.622204275747583,
|
| 6650 |
+
"grad_norm": 1.1812115907669067,
|
| 6651 |
+
"learning_rate": 1.9899886664148092e-05,
|
| 6652 |
+
"loss": 0.5395625114440918,
|
| 6653 |
+
"step": 9040
|
| 6654 |
+
},
|
| 6655 |
+
{
|
| 6656 |
+
"epoch": 1.6239989232115217,
|
| 6657 |
+
"grad_norm": 1.3128737211227417,
|
| 6658 |
+
"learning_rate": 1.9805440120891576e-05,
|
| 6659 |
+
"loss": 0.6198688983917237,
|
| 6660 |
+
"step": 9050
|
| 6661 |
+
},
|
| 6662 |
+
{
|
| 6663 |
+
"epoch": 1.6257935706754605,
|
| 6664 |
+
"grad_norm": 1.762397050857544,
|
| 6665 |
+
"learning_rate": 1.971099357763506e-05,
|
| 6666 |
+
"loss": 0.6085796356201172,
|
| 6667 |
+
"step": 9060
|
| 6668 |
+
},
|
| 6669 |
+
{
|
| 6670 |
+
"epoch": 1.6275882181393992,
|
| 6671 |
+
"grad_norm": 3.1758391857147217,
|
| 6672 |
+
"learning_rate": 1.9616547034378543e-05,
|
| 6673 |
+
"loss": 0.7153740882873535,
|
| 6674 |
+
"step": 9070
|
| 6675 |
+
},
|
| 6676 |
+
{
|
| 6677 |
+
"epoch": 1.629382865603338,
|
| 6678 |
+
"grad_norm": 0.8931203484535217,
|
| 6679 |
+
"learning_rate": 1.9522100491122026e-05,
|
| 6680 |
+
"loss": 0.7382801532745361,
|
| 6681 |
+
"step": 9080
|
| 6682 |
+
},
|
| 6683 |
+
{
|
| 6684 |
+
"epoch": 1.6311775130672768,
|
| 6685 |
+
"grad_norm": 1.3068408966064453,
|
| 6686 |
+
"learning_rate": 1.942765394786551e-05,
|
| 6687 |
+
"loss": 0.6057201862335205,
|
| 6688 |
+
"step": 9090
|
| 6689 |
+
},
|
| 6690 |
+
{
|
| 6691 |
+
"epoch": 1.6329721605312155,
|
| 6692 |
+
"grad_norm": 1.5849944353103638,
|
| 6693 |
+
"learning_rate": 1.933320740460899e-05,
|
| 6694 |
+
"loss": 0.5995753288269043,
|
| 6695 |
+
"step": 9100
|
| 6696 |
+
},
|
| 6697 |
+
{
|
| 6698 |
+
"epoch": 1.6347668079951545,
|
| 6699 |
+
"grad_norm": 1.6369379758834839,
|
| 6700 |
+
"learning_rate": 1.9238760861352476e-05,
|
| 6701 |
+
"loss": 0.6379215240478515,
|
| 6702 |
+
"step": 9110
|
| 6703 |
+
},
|
| 6704 |
+
{
|
| 6705 |
+
"epoch": 1.6365614554590933,
|
| 6706 |
+
"grad_norm": 2.743730306625366,
|
| 6707 |
+
"learning_rate": 1.914431431809596e-05,
|
| 6708 |
+
"loss": 0.6831157684326172,
|
| 6709 |
+
"step": 9120
|
| 6710 |
+
},
|
| 6711 |
+
{
|
| 6712 |
+
"epoch": 1.638356102923032,
|
| 6713 |
+
"grad_norm": 0.9385563731193542,
|
| 6714 |
+
"learning_rate": 1.9049867774839443e-05,
|
| 6715 |
+
"loss": 0.6559853076934814,
|
| 6716 |
+
"step": 9130
|
| 6717 |
+
},
|
| 6718 |
+
{
|
| 6719 |
+
"epoch": 1.640150750386971,
|
| 6720 |
+
"grad_norm": 1.053026556968689,
|
| 6721 |
+
"learning_rate": 1.8955421231582927e-05,
|
| 6722 |
+
"loss": 0.5699314594268798,
|
| 6723 |
+
"step": 9140
|
| 6724 |
+
},
|
| 6725 |
+
{
|
| 6726 |
+
"epoch": 1.6419453978509098,
|
| 6727 |
+
"grad_norm": 1.3851866722106934,
|
| 6728 |
+
"learning_rate": 1.8860974688326407e-05,
|
| 6729 |
+
"loss": 0.5622176170349121,
|
| 6730 |
+
"step": 9150
|
| 6731 |
+
},
|
| 6732 |
+
{
|
| 6733 |
+
"epoch": 1.6437400453148485,
|
| 6734 |
+
"grad_norm": 1.6981208324432373,
|
| 6735 |
+
"learning_rate": 1.8766528145069894e-05,
|
| 6736 |
+
"loss": 0.6120065212249756,
|
| 6737 |
+
"step": 9160
|
| 6738 |
+
},
|
| 6739 |
+
{
|
| 6740 |
+
"epoch": 1.6455346927787873,
|
| 6741 |
+
"grad_norm": 2.8609848022460938,
|
| 6742 |
+
"learning_rate": 1.8672081601813374e-05,
|
| 6743 |
+
"loss": 0.7159037590026855,
|
| 6744 |
+
"step": 9170
|
| 6745 |
+
},
|
| 6746 |
+
{
|
| 6747 |
+
"epoch": 1.647329340242726,
|
| 6748 |
+
"grad_norm": 0.8988924026489258,
|
| 6749 |
+
"learning_rate": 1.8577635058556857e-05,
|
| 6750 |
+
"loss": 0.5956008911132813,
|
| 6751 |
+
"step": 9180
|
| 6752 |
+
},
|
| 6753 |
+
{
|
| 6754 |
+
"epoch": 1.6491239877066648,
|
| 6755 |
+
"grad_norm": 1.1894454956054688,
|
| 6756 |
+
"learning_rate": 1.848318851530034e-05,
|
| 6757 |
+
"loss": 0.5765892028808594,
|
| 6758 |
+
"step": 9190
|
| 6759 |
+
},
|
| 6760 |
+
{
|
| 6761 |
+
"epoch": 1.6509186351706036,
|
| 6762 |
+
"grad_norm": 1.1688203811645508,
|
| 6763 |
+
"learning_rate": 1.8388741972043824e-05,
|
| 6764 |
+
"loss": 0.5537600040435791,
|
| 6765 |
+
"step": 9200
|
| 6766 |
+
},
|
| 6767 |
+
{
|
| 6768 |
+
"epoch": 1.6527132826345423,
|
| 6769 |
+
"grad_norm": 1.7588797807693481,
|
| 6770 |
+
"learning_rate": 1.8294295428787307e-05,
|
| 6771 |
+
"loss": 0.5942627906799316,
|
| 6772 |
+
"step": 9210
|
| 6773 |
+
},
|
| 6774 |
+
{
|
| 6775 |
+
"epoch": 1.6545079300984813,
|
| 6776 |
+
"grad_norm": 2.5389678478240967,
|
| 6777 |
+
"learning_rate": 1.819984888553079e-05,
|
| 6778 |
+
"loss": 0.6787733554840087,
|
| 6779 |
+
"step": 9220
|
| 6780 |
+
},
|
| 6781 |
+
{
|
| 6782 |
+
"epoch": 1.65630257756242,
|
| 6783 |
+
"grad_norm": 0.9558461904525757,
|
| 6784 |
+
"learning_rate": 1.8105402342274274e-05,
|
| 6785 |
+
"loss": 0.6737084865570069,
|
| 6786 |
+
"step": 9230
|
| 6787 |
+
},
|
| 6788 |
+
{
|
| 6789 |
+
"epoch": 1.658097225026359,
|
| 6790 |
+
"grad_norm": 1.2249739170074463,
|
| 6791 |
+
"learning_rate": 1.8010955799017757e-05,
|
| 6792 |
+
"loss": 0.5717909812927247,
|
| 6793 |
+
"step": 9240
|
| 6794 |
+
},
|
| 6795 |
+
{
|
| 6796 |
+
"epoch": 1.6598918724902978,
|
| 6797 |
+
"grad_norm": 1.2722245454788208,
|
| 6798 |
+
"learning_rate": 1.791650925576124e-05,
|
| 6799 |
+
"loss": 0.6410338878631592,
|
| 6800 |
+
"step": 9250
|
| 6801 |
+
},
|
| 6802 |
+
{
|
| 6803 |
+
"epoch": 1.6616865199542366,
|
| 6804 |
+
"grad_norm": 1.7197425365447998,
|
| 6805 |
+
"learning_rate": 1.782206271250472e-05,
|
| 6806 |
+
"loss": 0.5802127361297608,
|
| 6807 |
+
"step": 9260
|
| 6808 |
+
},
|
| 6809 |
+
{
|
| 6810 |
+
"epoch": 1.6634811674181753,
|
| 6811 |
+
"grad_norm": 2.545947551727295,
|
| 6812 |
+
"learning_rate": 1.7727616169248208e-05,
|
| 6813 |
+
"loss": 0.7030642986297607,
|
| 6814 |
+
"step": 9270
|
| 6815 |
+
},
|
| 6816 |
+
{
|
| 6817 |
+
"epoch": 1.665275814882114,
|
| 6818 |
+
"grad_norm": 0.9542912840843201,
|
| 6819 |
+
"learning_rate": 1.7633169625991688e-05,
|
| 6820 |
+
"loss": 0.7211085319519043,
|
| 6821 |
+
"step": 9280
|
| 6822 |
+
},
|
| 6823 |
+
{
|
| 6824 |
+
"epoch": 1.6670704623460528,
|
| 6825 |
+
"grad_norm": 1.1933677196502686,
|
| 6826 |
+
"learning_rate": 1.753872308273517e-05,
|
| 6827 |
+
"loss": 0.5878032207489013,
|
| 6828 |
+
"step": 9290
|
| 6829 |
+
},
|
| 6830 |
+
{
|
| 6831 |
+
"epoch": 1.6688651098099916,
|
| 6832 |
+
"grad_norm": 1.3516287803649902,
|
| 6833 |
+
"learning_rate": 1.7444276539478658e-05,
|
| 6834 |
+
"loss": 0.5172123432159423,
|
| 6835 |
+
"step": 9300
|
| 6836 |
+
},
|
| 6837 |
+
{
|
| 6838 |
+
"epoch": 1.6706597572739303,
|
| 6839 |
+
"grad_norm": 1.9860955476760864,
|
| 6840 |
+
"learning_rate": 1.7349829996222138e-05,
|
| 6841 |
+
"loss": 0.6105690002441406,
|
| 6842 |
+
"step": 9310
|
| 6843 |
+
},
|
| 6844 |
+
{
|
| 6845 |
+
"epoch": 1.6724544047378693,
|
| 6846 |
+
"grad_norm": 3.3884341716766357,
|
| 6847 |
+
"learning_rate": 1.7255383452965625e-05,
|
| 6848 |
+
"loss": 0.6953628063201904,
|
| 6849 |
+
"step": 9320
|
| 6850 |
+
},
|
| 6851 |
+
{
|
| 6852 |
+
"epoch": 1.674249052201808,
|
| 6853 |
+
"grad_norm": 1.0485398769378662,
|
| 6854 |
+
"learning_rate": 1.7160936909709105e-05,
|
| 6855 |
+
"loss": 0.6642786026000976,
|
| 6856 |
+
"step": 9330
|
| 6857 |
+
},
|
| 6858 |
+
{
|
| 6859 |
+
"epoch": 1.676043699665747,
|
| 6860 |
+
"grad_norm": 1.0807632207870483,
|
| 6861 |
+
"learning_rate": 1.706649036645259e-05,
|
| 6862 |
+
"loss": 0.5542885303497315,
|
| 6863 |
+
"step": 9340
|
| 6864 |
+
},
|
| 6865 |
+
{
|
| 6866 |
+
"epoch": 1.6778383471296858,
|
| 6867 |
+
"grad_norm": 1.3085391521453857,
|
| 6868 |
+
"learning_rate": 1.6972043823196072e-05,
|
| 6869 |
+
"loss": 0.5812755584716797,
|
| 6870 |
+
"step": 9350
|
| 6871 |
+
},
|
| 6872 |
+
{
|
| 6873 |
+
"epoch": 1.6796329945936246,
|
| 6874 |
+
"grad_norm": 1.8788141012191772,
|
| 6875 |
+
"learning_rate": 1.6877597279939555e-05,
|
| 6876 |
+
"loss": 0.5780276775360107,
|
| 6877 |
+
"step": 9360
|
| 6878 |
+
},
|
| 6879 |
+
{
|
| 6880 |
+
"epoch": 1.6814276420575633,
|
| 6881 |
+
"grad_norm": 2.884800434112549,
|
| 6882 |
+
"learning_rate": 1.678315073668304e-05,
|
| 6883 |
+
"loss": 0.6961386203765869,
|
| 6884 |
+
"step": 9370
|
| 6885 |
+
},
|
| 6886 |
+
{
|
| 6887 |
+
"epoch": 1.683222289521502,
|
| 6888 |
+
"grad_norm": 1.0055478811264038,
|
| 6889 |
+
"learning_rate": 1.6688704193426522e-05,
|
| 6890 |
+
"loss": 0.6600850582122803,
|
| 6891 |
+
"step": 9380
|
| 6892 |
+
},
|
| 6893 |
+
{
|
| 6894 |
+
"epoch": 1.6850169369854409,
|
| 6895 |
+
"grad_norm": 1.2578790187835693,
|
| 6896 |
+
"learning_rate": 1.6594257650170005e-05,
|
| 6897 |
+
"loss": 0.5850300788879395,
|
| 6898 |
+
"step": 9390
|
| 6899 |
+
},
|
| 6900 |
+
{
|
| 6901 |
+
"epoch": 1.6868115844493796,
|
| 6902 |
+
"grad_norm": 1.4355921745300293,
|
| 6903 |
+
"learning_rate": 1.649981110691349e-05,
|
| 6904 |
+
"loss": 0.5582671165466309,
|
| 6905 |
+
"step": 9400
|
| 6906 |
+
},
|
| 6907 |
+
{
|
| 6908 |
+
"epoch": 1.6886062319133184,
|
| 6909 |
+
"grad_norm": 1.8837885856628418,
|
| 6910 |
+
"learning_rate": 1.6405364563656972e-05,
|
| 6911 |
+
"loss": 0.5702266693115234,
|
| 6912 |
+
"step": 9410
|
| 6913 |
+
},
|
| 6914 |
+
{
|
| 6915 |
+
"epoch": 1.6904008793772574,
|
| 6916 |
+
"grad_norm": 2.5960443019866943,
|
| 6917 |
+
"learning_rate": 1.6310918020400452e-05,
|
| 6918 |
+
"loss": 0.7007398128509521,
|
| 6919 |
+
"step": 9420
|
| 6920 |
+
},
|
| 6921 |
+
{
|
| 6922 |
+
"epoch": 1.6921955268411961,
|
| 6923 |
+
"grad_norm": 1.0961920022964478,
|
| 6924 |
+
"learning_rate": 1.621647147714394e-05,
|
| 6925 |
+
"loss": 0.6672875881195068,
|
| 6926 |
+
"step": 9430
|
| 6927 |
+
},
|
| 6928 |
+
{
|
| 6929 |
+
"epoch": 1.693990174305135,
|
| 6930 |
+
"grad_norm": 1.3173032999038696,
|
| 6931 |
+
"learning_rate": 1.612202493388742e-05,
|
| 6932 |
+
"loss": 0.5785897731781006,
|
| 6933 |
+
"step": 9440
|
| 6934 |
+
},
|
| 6935 |
+
{
|
| 6936 |
+
"epoch": 1.6957848217690739,
|
| 6937 |
+
"grad_norm": 1.2982816696166992,
|
| 6938 |
+
"learning_rate": 1.6027578390630903e-05,
|
| 6939 |
+
"loss": 0.5776055335998536,
|
| 6940 |
+
"step": 9450
|
| 6941 |
+
},
|
| 6942 |
+
{
|
| 6943 |
+
"epoch": 1.6975794692330126,
|
| 6944 |
+
"grad_norm": 1.9547183513641357,
|
| 6945 |
+
"learning_rate": 1.5933131847374386e-05,
|
| 6946 |
+
"loss": 0.642759895324707,
|
| 6947 |
+
"step": 9460
|
| 6948 |
+
},
|
| 6949 |
+
{
|
| 6950 |
+
"epoch": 1.6993741166969514,
|
| 6951 |
+
"grad_norm": 3.1609091758728027,
|
| 6952 |
+
"learning_rate": 1.583868530411787e-05,
|
| 6953 |
+
"loss": 0.6947269916534424,
|
| 6954 |
+
"step": 9470
|
| 6955 |
+
},
|
| 6956 |
+
{
|
| 6957 |
+
"epoch": 1.7011687641608901,
|
| 6958 |
+
"grad_norm": 1.0354723930358887,
|
| 6959 |
+
"learning_rate": 1.5744238760861353e-05,
|
| 6960 |
+
"loss": 0.6658566474914551,
|
| 6961 |
+
"step": 9480
|
| 6962 |
+
},
|
| 6963 |
+
{
|
| 6964 |
+
"epoch": 1.702963411624829,
|
| 6965 |
+
"grad_norm": 1.160267949104309,
|
| 6966 |
+
"learning_rate": 1.5649792217604836e-05,
|
| 6967 |
+
"loss": 0.6192093372344971,
|
| 6968 |
+
"step": 9490
|
| 6969 |
+
},
|
| 6970 |
+
{
|
| 6971 |
+
"epoch": 1.7047580590887677,
|
| 6972 |
+
"grad_norm": 1.3688217401504517,
|
| 6973 |
+
"learning_rate": 1.555534567434832e-05,
|
| 6974 |
+
"loss": 0.5744320392608643,
|
| 6975 |
+
"step": 9500
|
| 6976 |
+
},
|
| 6977 |
+
{
|
| 6978 |
+
"epoch": 1.7065527065527064,
|
| 6979 |
+
"grad_norm": 1.8492395877838135,
|
| 6980 |
+
"learning_rate": 1.5460899131091803e-05,
|
| 6981 |
+
"loss": 0.647658109664917,
|
| 6982 |
+
"step": 9510
|
| 6983 |
+
},
|
| 6984 |
+
{
|
| 6985 |
+
"epoch": 1.7083473540166454,
|
| 6986 |
+
"grad_norm": 2.8793118000030518,
|
| 6987 |
+
"learning_rate": 1.5366452587835287e-05,
|
| 6988 |
+
"loss": 0.7834307193756104,
|
| 6989 |
+
"step": 9520
|
| 6990 |
+
},
|
| 6991 |
+
{
|
| 6992 |
+
"epoch": 1.7101420014805842,
|
| 6993 |
+
"grad_norm": 0.9082008600234985,
|
| 6994 |
+
"learning_rate": 1.527200604457877e-05,
|
| 6995 |
+
"loss": 0.6096126556396484,
|
| 6996 |
+
"step": 9530
|
| 6997 |
+
},
|
| 6998 |
+
{
|
| 6999 |
+
"epoch": 1.711936648944523,
|
| 7000 |
+
"grad_norm": 1.3358455896377563,
|
| 7001 |
+
"learning_rate": 1.5177559501322253e-05,
|
| 7002 |
+
"loss": 0.6171957015991211,
|
| 7003 |
+
"step": 9540
|
| 7004 |
+
},
|
| 7005 |
+
{
|
| 7006 |
+
"epoch": 1.713731296408462,
|
| 7007 |
+
"grad_norm": 1.6550923585891724,
|
| 7008 |
+
"learning_rate": 1.5083112958065735e-05,
|
| 7009 |
+
"loss": 0.5568831443786622,
|
| 7010 |
+
"step": 9550
|
| 7011 |
+
},
|
| 7012 |
+
{
|
| 7013 |
+
"epoch": 1.7155259438724006,
|
| 7014 |
+
"grad_norm": 1.9377288818359375,
|
| 7015 |
+
"learning_rate": 1.4988666414809219e-05,
|
| 7016 |
+
"loss": 0.6331680297851563,
|
| 7017 |
+
"step": 9560
|
| 7018 |
+
},
|
| 7019 |
+
{
|
| 7020 |
+
"epoch": 1.7173205913363394,
|
| 7021 |
+
"grad_norm": 3.0455734729766846,
|
| 7022 |
+
"learning_rate": 1.4894219871552704e-05,
|
| 7023 |
+
"loss": 0.6781876087188721,
|
| 7024 |
+
"step": 9570
|
| 7025 |
+
},
|
| 7026 |
+
{
|
| 7027 |
+
"epoch": 1.7191152388002782,
|
| 7028 |
+
"grad_norm": 0.8982117176055908,
|
| 7029 |
+
"learning_rate": 1.4799773328296185e-05,
|
| 7030 |
+
"loss": 0.6069835186004638,
|
| 7031 |
+
"step": 9580
|
| 7032 |
+
},
|
| 7033 |
+
{
|
| 7034 |
+
"epoch": 1.720909886264217,
|
| 7035 |
+
"grad_norm": 1.1043163537979126,
|
| 7036 |
+
"learning_rate": 1.4705326785039669e-05,
|
| 7037 |
+
"loss": 0.5600419044494629,
|
| 7038 |
+
"step": 9590
|
| 7039 |
+
},
|
| 7040 |
+
{
|
| 7041 |
+
"epoch": 1.7227045337281557,
|
| 7042 |
+
"grad_norm": 1.3668854236602783,
|
| 7043 |
+
"learning_rate": 1.461088024178315e-05,
|
| 7044 |
+
"loss": 0.559109878540039,
|
| 7045 |
+
"step": 9600
|
| 7046 |
+
},
|
| 7047 |
+
{
|
| 7048 |
+
"epoch": 1.7244991811920944,
|
| 7049 |
+
"grad_norm": 1.875893473625183,
|
| 7050 |
+
"learning_rate": 1.4516433698526636e-05,
|
| 7051 |
+
"loss": 0.5962122917175293,
|
| 7052 |
+
"step": 9610
|
| 7053 |
+
},
|
| 7054 |
+
{
|
| 7055 |
+
"epoch": 1.7262938286560334,
|
| 7056 |
+
"grad_norm": 2.8884074687957764,
|
| 7057 |
+
"learning_rate": 1.4421987155270117e-05,
|
| 7058 |
+
"loss": 0.7114584445953369,
|
| 7059 |
+
"step": 9620
|
| 7060 |
+
},
|
| 7061 |
+
{
|
| 7062 |
+
"epoch": 1.7280884761199722,
|
| 7063 |
+
"grad_norm": 1.1399059295654297,
|
| 7064 |
+
"learning_rate": 1.43275406120136e-05,
|
| 7065 |
+
"loss": 0.6676531791687011,
|
| 7066 |
+
"step": 9630
|
| 7067 |
+
},
|
| 7068 |
+
{
|
| 7069 |
+
"epoch": 1.729883123583911,
|
| 7070 |
+
"grad_norm": 1.235984444618225,
|
| 7071 |
+
"learning_rate": 1.4233094068757083e-05,
|
| 7072 |
+
"loss": 0.5513359546661377,
|
| 7073 |
+
"step": 9640
|
| 7074 |
+
},
|
| 7075 |
+
{
|
| 7076 |
+
"epoch": 1.73167777104785,
|
| 7077 |
+
"grad_norm": 1.3996710777282715,
|
| 7078 |
+
"learning_rate": 1.4138647525500568e-05,
|
| 7079 |
+
"loss": 0.5460638999938965,
|
| 7080 |
+
"step": 9650
|
| 7081 |
+
},
|
| 7082 |
+
{
|
| 7083 |
+
"epoch": 1.7334724185117887,
|
| 7084 |
+
"grad_norm": 1.8478695154190063,
|
| 7085 |
+
"learning_rate": 1.4044200982244051e-05,
|
| 7086 |
+
"loss": 0.6401609897613525,
|
| 7087 |
+
"step": 9660
|
| 7088 |
+
},
|
| 7089 |
+
{
|
| 7090 |
+
"epoch": 1.7352670659757274,
|
| 7091 |
+
"grad_norm": 3.396739959716797,
|
| 7092 |
+
"learning_rate": 1.3949754438987533e-05,
|
| 7093 |
+
"loss": 0.6967354774475097,
|
| 7094 |
+
"step": 9670
|
| 7095 |
+
},
|
| 7096 |
+
{
|
| 7097 |
+
"epoch": 1.7370617134396662,
|
| 7098 |
+
"grad_norm": 1.0562396049499512,
|
| 7099 |
+
"learning_rate": 1.3855307895731018e-05,
|
| 7100 |
+
"loss": 0.6651877403259278,
|
| 7101 |
+
"step": 9680
|
| 7102 |
+
},
|
| 7103 |
+
{
|
| 7104 |
+
"epoch": 1.738856360903605,
|
| 7105 |
+
"grad_norm": 1.1792099475860596,
|
| 7106 |
+
"learning_rate": 1.37608613524745e-05,
|
| 7107 |
+
"loss": 0.5408987998962402,
|
| 7108 |
+
"step": 9690
|
| 7109 |
+
},
|
| 7110 |
+
{
|
| 7111 |
+
"epoch": 1.7406510083675437,
|
| 7112 |
+
"grad_norm": 1.4664831161499023,
|
| 7113 |
+
"learning_rate": 1.3666414809217985e-05,
|
| 7114 |
+
"loss": 0.5653331756591797,
|
| 7115 |
+
"step": 9700
|
| 7116 |
+
},
|
| 7117 |
+
{
|
| 7118 |
+
"epoch": 1.7424456558314825,
|
| 7119 |
+
"grad_norm": 1.6036666631698608,
|
| 7120 |
+
"learning_rate": 1.3571968265961465e-05,
|
| 7121 |
+
"loss": 0.625031566619873,
|
| 7122 |
+
"step": 9710
|
| 7123 |
+
},
|
| 7124 |
+
{
|
| 7125 |
+
"epoch": 1.7442403032954215,
|
| 7126 |
+
"grad_norm": 2.804121255874634,
|
| 7127 |
+
"learning_rate": 1.347752172270495e-05,
|
| 7128 |
+
"loss": 0.6458487987518311,
|
| 7129 |
+
"step": 9720
|
| 7130 |
+
},
|
| 7131 |
+
{
|
| 7132 |
+
"epoch": 1.7460349507593602,
|
| 7133 |
+
"grad_norm": 0.9326400756835938,
|
| 7134 |
+
"learning_rate": 1.3383075179448432e-05,
|
| 7135 |
+
"loss": 0.6127943515777587,
|
| 7136 |
+
"step": 9730
|
| 7137 |
+
},
|
| 7138 |
+
{
|
| 7139 |
+
"epoch": 1.747829598223299,
|
| 7140 |
+
"grad_norm": 1.133003830909729,
|
| 7141 |
+
"learning_rate": 1.3288628636191917e-05,
|
| 7142 |
+
"loss": 0.5575729846954346,
|
| 7143 |
+
"step": 9740
|
| 7144 |
+
},
|
| 7145 |
+
{
|
| 7146 |
+
"epoch": 1.749624245687238,
|
| 7147 |
+
"grad_norm": 1.4575793743133545,
|
| 7148 |
+
"learning_rate": 1.31941820929354e-05,
|
| 7149 |
+
"loss": 0.576661491394043,
|
| 7150 |
+
"step": 9750
|
| 7151 |
+
},
|
| 7152 |
+
{
|
| 7153 |
+
"epoch": 1.7514188931511767,
|
| 7154 |
+
"grad_norm": 1.9434808492660522,
|
| 7155 |
+
"learning_rate": 1.3099735549678882e-05,
|
| 7156 |
+
"loss": 0.6073297977447509,
|
| 7157 |
+
"step": 9760
|
| 7158 |
+
},
|
| 7159 |
+
{
|
| 7160 |
+
"epoch": 1.7532135406151155,
|
| 7161 |
+
"grad_norm": 3.4238226413726807,
|
| 7162 |
+
"learning_rate": 1.3005289006422367e-05,
|
| 7163 |
+
"loss": 0.6530107498168946,
|
| 7164 |
+
"step": 9770
|
| 7165 |
+
},
|
| 7166 |
+
{
|
| 7167 |
+
"epoch": 1.7550081880790542,
|
| 7168 |
+
"grad_norm": 0.948078989982605,
|
| 7169 |
+
"learning_rate": 1.2910842463165849e-05,
|
| 7170 |
+
"loss": 0.706777811050415,
|
| 7171 |
+
"step": 9780
|
| 7172 |
+
},
|
| 7173 |
+
{
|
| 7174 |
+
"epoch": 1.756802835542993,
|
| 7175 |
+
"grad_norm": 1.269318699836731,
|
| 7176 |
+
"learning_rate": 1.2816395919909332e-05,
|
| 7177 |
+
"loss": 0.5674312114715576,
|
| 7178 |
+
"step": 9790
|
| 7179 |
+
},
|
| 7180 |
+
{
|
| 7181 |
+
"epoch": 1.7585974830069317,
|
| 7182 |
+
"grad_norm": 1.5564656257629395,
|
| 7183 |
+
"learning_rate": 1.2721949376652814e-05,
|
| 7184 |
+
"loss": 0.6226547241210938,
|
| 7185 |
+
"step": 9800
|
| 7186 |
+
},
|
| 7187 |
+
{
|
| 7188 |
+
"epoch": 1.7603921304708705,
|
| 7189 |
+
"grad_norm": 1.7986268997192383,
|
| 7190 |
+
"learning_rate": 1.2627502833396299e-05,
|
| 7191 |
+
"loss": 0.6185244083404541,
|
| 7192 |
+
"step": 9810
|
| 7193 |
+
},
|
| 7194 |
+
{
|
| 7195 |
+
"epoch": 1.7621867779348093,
|
| 7196 |
+
"grad_norm": 2.82063889503479,
|
| 7197 |
+
"learning_rate": 1.253305629013978e-05,
|
| 7198 |
+
"loss": 0.6908327579498291,
|
| 7199 |
+
"step": 9820
|
| 7200 |
+
},
|
| 7201 |
+
{
|
| 7202 |
+
"epoch": 1.7639814253987482,
|
| 7203 |
+
"grad_norm": 0.9118233919143677,
|
| 7204 |
+
"learning_rate": 1.2438609746883264e-05,
|
| 7205 |
+
"loss": 0.6583641052246094,
|
| 7206 |
+
"step": 9830
|
| 7207 |
+
},
|
| 7208 |
+
{
|
| 7209 |
+
"epoch": 1.765776072862687,
|
| 7210 |
+
"grad_norm": 1.270808458328247,
|
| 7211 |
+
"learning_rate": 1.2344163203626748e-05,
|
| 7212 |
+
"loss": 0.6372414112091065,
|
| 7213 |
+
"step": 9840
|
| 7214 |
+
},
|
| 7215 |
+
{
|
| 7216 |
+
"epoch": 1.767570720326626,
|
| 7217 |
+
"grad_norm": 1.4152858257293701,
|
| 7218 |
+
"learning_rate": 1.2249716660370231e-05,
|
| 7219 |
+
"loss": 0.5774289131164551,
|
| 7220 |
+
"step": 9850
|
| 7221 |
+
},
|
| 7222 |
+
{
|
| 7223 |
+
"epoch": 1.7693653677905647,
|
| 7224 |
+
"grad_norm": 1.7652438879013062,
|
| 7225 |
+
"learning_rate": 1.2155270117113714e-05,
|
| 7226 |
+
"loss": 0.576290512084961,
|
| 7227 |
+
"step": 9860
|
| 7228 |
+
},
|
| 7229 |
+
{
|
| 7230 |
+
"epoch": 1.7711600152545035,
|
| 7231 |
+
"grad_norm": 2.593611478805542,
|
| 7232 |
+
"learning_rate": 1.2060823573857196e-05,
|
| 7233 |
+
"loss": 0.5927243709564209,
|
| 7234 |
+
"step": 9870
|
| 7235 |
+
},
|
| 7236 |
+
{
|
| 7237 |
+
"epoch": 1.7729546627184423,
|
| 7238 |
+
"grad_norm": 1.003075361251831,
|
| 7239 |
+
"learning_rate": 1.1966377030600681e-05,
|
| 7240 |
+
"loss": 0.6748008251190185,
|
| 7241 |
+
"step": 9880
|
| 7242 |
+
},
|
| 7243 |
+
{
|
| 7244 |
+
"epoch": 1.774749310182381,
|
| 7245 |
+
"grad_norm": 0.9902793765068054,
|
| 7246 |
+
"learning_rate": 1.1871930487344165e-05,
|
| 7247 |
+
"loss": 0.5589398384094239,
|
| 7248 |
+
"step": 9890
|
| 7249 |
+
},
|
| 7250 |
+
{
|
| 7251 |
+
"epoch": 1.7765439576463198,
|
| 7252 |
+
"grad_norm": 1.4090533256530762,
|
| 7253 |
+
"learning_rate": 1.1777483944087646e-05,
|
| 7254 |
+
"loss": 0.5819490432739258,
|
| 7255 |
+
"step": 9900
|
| 7256 |
+
},
|
| 7257 |
+
{
|
| 7258 |
+
"epoch": 1.7783386051102585,
|
| 7259 |
+
"grad_norm": 2.1041059494018555,
|
| 7260 |
+
"learning_rate": 1.168303740083113e-05,
|
| 7261 |
+
"loss": 0.6257413864135742,
|
| 7262 |
+
"step": 9910
|
| 7263 |
+
},
|
| 7264 |
+
{
|
| 7265 |
+
"epoch": 1.7801332525741973,
|
| 7266 |
+
"grad_norm": 3.1887195110321045,
|
| 7267 |
+
"learning_rate": 1.1588590857574613e-05,
|
| 7268 |
+
"loss": 0.6508440017700196,
|
| 7269 |
+
"step": 9920
|
| 7270 |
+
},
|
| 7271 |
+
{
|
| 7272 |
+
"epoch": 1.7819279000381363,
|
| 7273 |
+
"grad_norm": 1.0658595561981201,
|
| 7274 |
+
"learning_rate": 1.1494144314318097e-05,
|
| 7275 |
+
"loss": 0.6383107662200928,
|
| 7276 |
+
"step": 9930
|
| 7277 |
+
},
|
| 7278 |
+
{
|
| 7279 |
+
"epoch": 1.783722547502075,
|
| 7280 |
+
"grad_norm": 1.3658292293548584,
|
| 7281 |
+
"learning_rate": 1.139969777106158e-05,
|
| 7282 |
+
"loss": 0.6108718395233155,
|
| 7283 |
+
"step": 9940
|
| 7284 |
+
},
|
| 7285 |
+
{
|
| 7286 |
+
"epoch": 1.785517194966014,
|
| 7287 |
+
"grad_norm": 1.6356791257858276,
|
| 7288 |
+
"learning_rate": 1.1305251227805062e-05,
|
| 7289 |
+
"loss": 0.5537287712097168,
|
| 7290 |
+
"step": 9950
|
| 7291 |
+
},
|
| 7292 |
+
{
|
| 7293 |
+
"epoch": 1.7873118424299528,
|
| 7294 |
+
"grad_norm": 1.9489909410476685,
|
| 7295 |
+
"learning_rate": 1.1210804684548545e-05,
|
| 7296 |
+
"loss": 0.6689440250396729,
|
| 7297 |
+
"step": 9960
|
| 7298 |
+
},
|
| 7299 |
+
{
|
| 7300 |
+
"epoch": 1.7891064898938915,
|
| 7301 |
+
"grad_norm": 3.880660057067871,
|
| 7302 |
+
"learning_rate": 1.111635814129203e-05,
|
| 7303 |
+
"loss": 0.7151643753051757,
|
| 7304 |
+
"step": 9970
|
| 7305 |
+
},
|
| 7306 |
+
{
|
| 7307 |
+
"epoch": 1.7909011373578303,
|
| 7308 |
+
"grad_norm": 0.9832514524459839,
|
| 7309 |
+
"learning_rate": 1.1021911598035512e-05,
|
| 7310 |
+
"loss": 0.7017166137695312,
|
| 7311 |
+
"step": 9980
|
| 7312 |
+
},
|
| 7313 |
+
{
|
| 7314 |
+
"epoch": 1.792695784821769,
|
| 7315 |
+
"grad_norm": 1.2020207643508911,
|
| 7316 |
+
"learning_rate": 1.0927465054778996e-05,
|
| 7317 |
+
"loss": 0.5730813026428223,
|
| 7318 |
+
"step": 9990
|
| 7319 |
+
},
|
| 7320 |
+
{
|
| 7321 |
+
"epoch": 1.7944904322857078,
|
| 7322 |
+
"grad_norm": 1.3736741542816162,
|
| 7323 |
+
"learning_rate": 1.0833018511522479e-05,
|
| 7324 |
+
"loss": 0.5724750995635987,
|
| 7325 |
+
"step": 10000
|
| 7326 |
+
},
|
| 7327 |
+
{
|
| 7328 |
+
"epoch": 1.7944904322857078,
|
| 7329 |
+
"step": 10000,
|
| 7330 |
+
"test_entity_exact_match": 0.0029702970297029703,
|
| 7331 |
+
"test_entity_fix_rate": 0.24977299057519492,
|
| 7332 |
+
"test_entity_fix_rate_lenient": 0.24275862068965517,
|
| 7333 |
+
"test_entity_fix_rate_recoverable": 0.24613225515420858,
|
| 7334 |
+
"test_entity_hallucination_rate": 0.0039603960396039604,
|
| 7335 |
+
"test_entity_preservation_rate": 0.7072993405179347,
|
| 7336 |
+
"test_entity_preservation_rate_lenient": 0.7093837804549479,
|
| 7337 |
+
"test_entity_targeted_score": 0.47853616554656486,
|
| 7338 |
+
"test_entity_targeted_score_lenient": 0.4760712005723015,
|
| 7339 |
+
"test_entity_wer": 0.3921019203467085,
|
| 7340 |
+
"test_entity_wer_zwnj_normalized": 0.38817369584013833
|
| 7341 |
+
},
|
| 7342 |
+
{
|
| 7343 |
+
"epoch": 1.7944904322857078,
|
| 7344 |
+
"step": 10000,
|
| 7345 |
+
"test_typo_exact_match": 0.013966480446927373,
|
| 7346 |
+
"test_typo_fix_rate": 0.3449011244668476,
|
| 7347 |
+
"test_typo_fix_rate_lenient": 0.33719726764645,
|
| 7348 |
+
"test_typo_fix_rate_recoverable": 0.34153005464480873,
|
| 7349 |
+
"test_typo_hallucination_rate": 0.019553072625698324,
|
| 7350 |
+
"test_typo_preservation_rate": 0.8174227501249522,
|
| 7351 |
+
"test_typo_preservation_rate_lenient": 0.8196255383984294,
|
| 7352 |
+
"test_typo_targeted_score": 0.5811619372958999,
|
| 7353 |
+
"test_typo_targeted_score_lenient": 0.5784114030224398,
|
| 7354 |
+
"test_typo_wer": 0.2880191978759797,
|
| 7355 |
+
"test_typo_wer_zwnj_normalized": 0.2832963161522555
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 1.7944904322857078,
|
| 7359 |
+
"step": 10000,
|
| 7360 |
+
"test_best_wer": 0.2231321104545041
|
| 7361 |
+
},
|
| 7362 |
+
{
|
| 7363 |
+
"epoch": 1.7962850797496466,
|
| 7364 |
+
"grad_norm": 2.1501452922821045,
|
| 7365 |
+
"learning_rate": 1.0738571968265962e-05,
|
| 7366 |
+
"loss": 0.5956575393676757,
|
| 7367 |
+
"step": 10010
|
| 7368 |
+
},
|
| 7369 |
+
{
|
| 7370 |
+
"epoch": 1.7980797272135853,
|
| 7371 |
+
"grad_norm": 3.254652738571167,
|
| 7372 |
+
"learning_rate": 1.0644125425009444e-05,
|
| 7373 |
+
"loss": 0.6456453800201416,
|
| 7374 |
+
"step": 10020
|
| 7375 |
+
},
|
| 7376 |
+
{
|
| 7377 |
+
"epoch": 1.7998743746775243,
|
| 7378 |
+
"grad_norm": 0.951464056968689,
|
| 7379 |
+
"learning_rate": 1.0549678881752928e-05,
|
| 7380 |
+
"loss": 0.7187893390655518,
|
| 7381 |
+
"step": 10030
|
| 7382 |
+
},
|
| 7383 |
+
{
|
| 7384 |
+
"epoch": 1.801669022141463,
|
| 7385 |
+
"grad_norm": 1.395233392715454,
|
| 7386 |
+
"learning_rate": 1.0455232338496411e-05,
|
| 7387 |
+
"loss": 0.590334701538086,
|
| 7388 |
+
"step": 10040
|
| 7389 |
+
},
|
| 7390 |
+
{
|
| 7391 |
+
"epoch": 1.803463669605402,
|
| 7392 |
+
"grad_norm": 1.5860689878463745,
|
| 7393 |
+
"learning_rate": 1.0360785795239896e-05,
|
| 7394 |
+
"loss": 0.5618289470672607,
|
| 7395 |
+
"step": 10050
|
| 7396 |
+
},
|
| 7397 |
+
{
|
| 7398 |
+
"epoch": 1.8052583170693408,
|
| 7399 |
+
"grad_norm": 1.9702314138412476,
|
| 7400 |
+
"learning_rate": 1.0266339251983378e-05,
|
| 7401 |
+
"loss": 0.5563257217407227,
|
| 7402 |
+
"step": 10060
|
| 7403 |
+
},
|
| 7404 |
+
{
|
| 7405 |
+
"epoch": 1.8070529645332796,
|
| 7406 |
+
"grad_norm": 2.8797783851623535,
|
| 7407 |
+
"learning_rate": 1.0171892708726861e-05,
|
| 7408 |
+
"loss": 0.5672471046447753,
|
| 7409 |
+
"step": 10070
|
| 7410 |
+
},
|
| 7411 |
+
{
|
| 7412 |
+
"epoch": 1.8088476119972183,
|
| 7413 |
+
"grad_norm": 0.9113657474517822,
|
| 7414 |
+
"learning_rate": 1.0077446165470345e-05,
|
| 7415 |
+
"loss": 0.6514238834381103,
|
| 7416 |
+
"step": 10080
|
| 7417 |
+
},
|
| 7418 |
+
{
|
| 7419 |
+
"epoch": 1.810642259461157,
|
| 7420 |
+
"grad_norm": 1.2037993669509888,
|
| 7421 |
+
"learning_rate": 9.982999622213828e-06,
|
| 7422 |
+
"loss": 0.5554754734039307,
|
| 7423 |
+
"step": 10090
|
| 7424 |
+
},
|
| 7425 |
+
{
|
| 7426 |
+
"epoch": 1.8124369069250958,
|
| 7427 |
+
"grad_norm": 1.5878387689590454,
|
| 7428 |
+
"learning_rate": 9.88855307895731e-06,
|
| 7429 |
+
"loss": 0.5513728141784668,
|
| 7430 |
+
"step": 10100
|
| 7431 |
+
},
|
| 7432 |
+
{
|
| 7433 |
+
"epoch": 1.8142315543890346,
|
| 7434 |
+
"grad_norm": 1.9183306694030762,
|
| 7435 |
+
"learning_rate": 9.794106535700793e-06,
|
| 7436 |
+
"loss": 0.5588463306427002,
|
| 7437 |
+
"step": 10110
|
| 7438 |
+
},
|
| 7439 |
+
{
|
| 7440 |
+
"epoch": 1.8160262018529734,
|
| 7441 |
+
"grad_norm": 2.983146905899048,
|
| 7442 |
+
"learning_rate": 9.699659992444277e-06,
|
| 7443 |
+
"loss": 0.7038702487945556,
|
| 7444 |
+
"step": 10120
|
| 7445 |
+
},
|
| 7446 |
+
{
|
| 7447 |
+
"epoch": 1.8178208493169123,
|
| 7448 |
+
"grad_norm": 0.9136666059494019,
|
| 7449 |
+
"learning_rate": 9.60521344918776e-06,
|
| 7450 |
+
"loss": 0.6319262027740479,
|
| 7451 |
+
"step": 10130
|
| 7452 |
+
},
|
| 7453 |
+
{
|
| 7454 |
+
"epoch": 1.819615496780851,
|
| 7455 |
+
"grad_norm": 1.3056459426879883,
|
| 7456 |
+
"learning_rate": 9.510766905931243e-06,
|
| 7457 |
+
"loss": 0.6007857322692871,
|
| 7458 |
+
"step": 10140
|
| 7459 |
+
},
|
| 7460 |
+
{
|
| 7461 |
+
"epoch": 1.8214101442447899,
|
| 7462 |
+
"grad_norm": 1.5381118059158325,
|
| 7463 |
+
"learning_rate": 9.416320362674727e-06,
|
| 7464 |
+
"loss": 0.5293020248413086,
|
| 7465 |
+
"step": 10150
|
| 7466 |
+
},
|
| 7467 |
+
{
|
| 7468 |
+
"epoch": 1.8232047917087288,
|
| 7469 |
+
"grad_norm": 1.9120346307754517,
|
| 7470 |
+
"learning_rate": 9.32187381941821e-06,
|
| 7471 |
+
"loss": 0.581623125076294,
|
| 7472 |
+
"step": 10160
|
| 7473 |
+
},
|
| 7474 |
+
{
|
| 7475 |
+
"epoch": 1.8249994391726676,
|
| 7476 |
+
"grad_norm": 2.7101519107818604,
|
| 7477 |
+
"learning_rate": 9.227427276161694e-06,
|
| 7478 |
+
"loss": 0.7141891479492187,
|
| 7479 |
+
"step": 10170
|
| 7480 |
+
},
|
| 7481 |
+
{
|
| 7482 |
+
"epoch": 1.8267940866366064,
|
| 7483 |
+
"grad_norm": 1.0163861513137817,
|
| 7484 |
+
"learning_rate": 9.132980732905175e-06,
|
| 7485 |
+
"loss": 0.6416571617126465,
|
| 7486 |
+
"step": 10180
|
| 7487 |
+
},
|
| 7488 |
+
{
|
| 7489 |
+
"epoch": 1.8285887341005451,
|
| 7490 |
+
"grad_norm": 1.0968486070632935,
|
| 7491 |
+
"learning_rate": 9.038534189648659e-06,
|
| 7492 |
+
"loss": 0.5787481307983399,
|
| 7493 |
+
"step": 10190
|
| 7494 |
+
},
|
| 7495 |
+
{
|
| 7496 |
+
"epoch": 1.8303833815644839,
|
| 7497 |
+
"grad_norm": 1.5518709421157837,
|
| 7498 |
+
"learning_rate": 8.944087646392142e-06,
|
| 7499 |
+
"loss": 0.5948922157287597,
|
| 7500 |
+
"step": 10200
|
| 7501 |
+
},
|
| 7502 |
+
{
|
| 7503 |
+
"epoch": 1.8321780290284226,
|
| 7504 |
+
"grad_norm": 1.6701586246490479,
|
| 7505 |
+
"learning_rate": 8.849641103135626e-06,
|
| 7506 |
+
"loss": 0.6065248489379883,
|
| 7507 |
+
"step": 10210
|
| 7508 |
+
},
|
| 7509 |
+
{
|
| 7510 |
+
"epoch": 1.8339726764923614,
|
| 7511 |
+
"grad_norm": 2.888354539871216,
|
| 7512 |
+
"learning_rate": 8.755194559879107e-06,
|
| 7513 |
+
"loss": 0.6854756832122803,
|
| 7514 |
+
"step": 10220
|
| 7515 |
+
},
|
| 7516 |
+
{
|
| 7517 |
+
"epoch": 1.8357673239563004,
|
| 7518 |
+
"grad_norm": 1.063319444656372,
|
| 7519 |
+
"learning_rate": 8.660748016622593e-06,
|
| 7520 |
+
"loss": 0.6407892227172851,
|
| 7521 |
+
"step": 10230
|
| 7522 |
+
},
|
| 7523 |
+
{
|
| 7524 |
+
"epoch": 1.8375619714202391,
|
| 7525 |
+
"grad_norm": 1.2017849683761597,
|
| 7526 |
+
"learning_rate": 8.566301473366076e-06,
|
| 7527 |
+
"loss": 0.5319945812225342,
|
| 7528 |
+
"step": 10240
|
| 7529 |
+
},
|
| 7530 |
+
{
|
| 7531 |
+
"epoch": 1.839356618884178,
|
| 7532 |
+
"grad_norm": 1.6157032251358032,
|
| 7533 |
+
"learning_rate": 8.47185493010956e-06,
|
| 7534 |
+
"loss": 0.5466680526733398,
|
| 7535 |
+
"step": 10250
|
| 7536 |
+
},
|
| 7537 |
+
{
|
| 7538 |
+
"epoch": 1.8411512663481169,
|
| 7539 |
+
"grad_norm": 1.9785637855529785,
|
| 7540 |
+
"learning_rate": 8.377408386853041e-06,
|
| 7541 |
+
"loss": 0.5784874439239502,
|
| 7542 |
+
"step": 10260
|
| 7543 |
+
},
|
| 7544 |
+
{
|
| 7545 |
+
"epoch": 1.8429459138120556,
|
| 7546 |
+
"grad_norm": 3.0544188022613525,
|
| 7547 |
+
"learning_rate": 8.282961843596525e-06,
|
| 7548 |
+
"loss": 0.7072327136993408,
|
| 7549 |
+
"step": 10270
|
| 7550 |
+
},
|
| 7551 |
+
{
|
| 7552 |
+
"epoch": 1.8447405612759944,
|
| 7553 |
+
"grad_norm": 1.02102792263031,
|
| 7554 |
+
"learning_rate": 8.188515300340008e-06,
|
| 7555 |
+
"loss": 0.6501930236816407,
|
| 7556 |
+
"step": 10280
|
| 7557 |
+
},
|
| 7558 |
+
{
|
| 7559 |
+
"epoch": 1.8465352087399332,
|
| 7560 |
+
"grad_norm": 1.2776883840560913,
|
| 7561 |
+
"learning_rate": 8.094068757083491e-06,
|
| 7562 |
+
"loss": 0.5501949787139893,
|
| 7563 |
+
"step": 10290
|
| 7564 |
+
},
|
| 7565 |
+
{
|
| 7566 |
+
"epoch": 1.848329856203872,
|
| 7567 |
+
"grad_norm": 1.5707981586456299,
|
| 7568 |
+
"learning_rate": 7.999622213826973e-06,
|
| 7569 |
+
"loss": 0.5422088146209717,
|
| 7570 |
+
"step": 10300
|
| 7571 |
+
},
|
| 7572 |
+
{
|
| 7573 |
+
"epoch": 1.8501245036678107,
|
| 7574 |
+
"grad_norm": 1.9141781330108643,
|
| 7575 |
+
"learning_rate": 7.905175670570457e-06,
|
| 7576 |
+
"loss": 0.6311410427093506,
|
| 7577 |
+
"step": 10310
|
| 7578 |
+
},
|
| 7579 |
+
{
|
| 7580 |
+
"epoch": 1.8519191511317494,
|
| 7581 |
+
"grad_norm": 2.775017499923706,
|
| 7582 |
+
"learning_rate": 7.810729127313942e-06,
|
| 7583 |
+
"loss": 0.6945941925048829,
|
| 7584 |
+
"step": 10320
|
| 7585 |
+
},
|
| 7586 |
+
{
|
| 7587 |
+
"epoch": 1.8537137985956884,
|
| 7588 |
+
"grad_norm": 0.9610602855682373,
|
| 7589 |
+
"learning_rate": 7.716282584057425e-06,
|
| 7590 |
+
"loss": 0.6370384216308593,
|
| 7591 |
+
"step": 10330
|
| 7592 |
+
},
|
| 7593 |
+
{
|
| 7594 |
+
"epoch": 1.8555084460596272,
|
| 7595 |
+
"grad_norm": 1.2685805559158325,
|
| 7596 |
+
"learning_rate": 7.621836040800908e-06,
|
| 7597 |
+
"loss": 0.5275514602661133,
|
| 7598 |
+
"step": 10340
|
| 7599 |
+
},
|
| 7600 |
+
{
|
| 7601 |
+
"epoch": 1.857303093523566,
|
| 7602 |
+
"grad_norm": 1.606128215789795,
|
| 7603 |
+
"learning_rate": 7.52738949754439e-06,
|
| 7604 |
+
"loss": 0.5831241130828857,
|
| 7605 |
+
"step": 10350
|
| 7606 |
+
},
|
| 7607 |
+
{
|
| 7608 |
+
"epoch": 1.859097740987505,
|
| 7609 |
+
"grad_norm": 1.875279426574707,
|
| 7610 |
+
"learning_rate": 7.432942954287874e-06,
|
| 7611 |
+
"loss": 0.574491024017334,
|
| 7612 |
+
"step": 10360
|
| 7613 |
+
},
|
| 7614 |
+
{
|
| 7615 |
+
"epoch": 1.8608923884514437,
|
| 7616 |
+
"grad_norm": 2.2604353427886963,
|
| 7617 |
+
"learning_rate": 7.338496411031356e-06,
|
| 7618 |
+
"loss": 0.6062031269073487,
|
| 7619 |
+
"step": 10370
|
| 7620 |
+
},
|
| 7621 |
+
{
|
| 7622 |
+
"epoch": 1.8626870359153824,
|
| 7623 |
+
"grad_norm": 0.9294455051422119,
|
| 7624 |
+
"learning_rate": 7.24404986777484e-06,
|
| 7625 |
+
"loss": 0.6486630439758301,
|
| 7626 |
+
"step": 10380
|
| 7627 |
+
},
|
| 7628 |
+
{
|
| 7629 |
+
"epoch": 1.8644816833793212,
|
| 7630 |
+
"grad_norm": 1.2625101804733276,
|
| 7631 |
+
"learning_rate": 7.149603324518322e-06,
|
| 7632 |
+
"loss": 0.5553452014923096,
|
| 7633 |
+
"step": 10390
|
| 7634 |
+
},
|
| 7635 |
+
{
|
| 7636 |
+
"epoch": 1.86627633084326,
|
| 7637 |
+
"grad_norm": 1.5612289905548096,
|
| 7638 |
+
"learning_rate": 7.055156781261806e-06,
|
| 7639 |
+
"loss": 0.5661460399627686,
|
| 7640 |
+
"step": 10400
|
| 7641 |
+
},
|
| 7642 |
+
{
|
| 7643 |
+
"epoch": 1.8680709783071987,
|
| 7644 |
+
"grad_norm": 1.7258232831954956,
|
| 7645 |
+
"learning_rate": 6.96071023800529e-06,
|
| 7646 |
+
"loss": 0.5621381759643554,
|
| 7647 |
+
"step": 10410
|
| 7648 |
+
},
|
| 7649 |
+
{
|
| 7650 |
+
"epoch": 1.8698656257711375,
|
| 7651 |
+
"grad_norm": 3.043125629425049,
|
| 7652 |
+
"learning_rate": 6.866263694748773e-06,
|
| 7653 |
+
"loss": 0.6614994525909423,
|
| 7654 |
+
"step": 10420
|
| 7655 |
+
},
|
| 7656 |
+
{
|
| 7657 |
+
"epoch": 1.8716602732350762,
|
| 7658 |
+
"grad_norm": 1.081845998764038,
|
| 7659 |
+
"learning_rate": 6.771817151492256e-06,
|
| 7660 |
+
"loss": 0.6136946678161621,
|
| 7661 |
+
"step": 10430
|
| 7662 |
+
},
|
| 7663 |
+
{
|
| 7664 |
+
"epoch": 1.8734549206990152,
|
| 7665 |
+
"grad_norm": 1.091383457183838,
|
| 7666 |
+
"learning_rate": 6.677370608235739e-06,
|
| 7667 |
+
"loss": 0.5470185279846191,
|
| 7668 |
+
"step": 10440
|
| 7669 |
+
},
|
| 7670 |
+
{
|
| 7671 |
+
"epoch": 1.875249568162954,
|
| 7672 |
+
"grad_norm": 1.5916881561279297,
|
| 7673 |
+
"learning_rate": 6.582924064979222e-06,
|
| 7674 |
+
"loss": 0.6097097873687745,
|
| 7675 |
+
"step": 10450
|
| 7676 |
+
},
|
| 7677 |
+
{
|
| 7678 |
+
"epoch": 1.877044215626893,
|
| 7679 |
+
"grad_norm": 1.6788287162780762,
|
| 7680 |
+
"learning_rate": 6.488477521722705e-06,
|
| 7681 |
+
"loss": 0.5724509716033935,
|
| 7682 |
+
"step": 10460
|
| 7683 |
+
},
|
| 7684 |
+
{
|
| 7685 |
+
"epoch": 1.8788388630908317,
|
| 7686 |
+
"grad_norm": 3.0279879570007324,
|
| 7687 |
+
"learning_rate": 6.394030978466188e-06,
|
| 7688 |
+
"loss": 0.7021475791931152,
|
| 7689 |
+
"step": 10470
|
| 7690 |
+
},
|
| 7691 |
+
{
|
| 7692 |
+
"epoch": 1.8806335105547705,
|
| 7693 |
+
"grad_norm": 0.9188269972801208,
|
| 7694 |
+
"learning_rate": 6.299584435209671e-06,
|
| 7695 |
+
"loss": 0.6713749408721924,
|
| 7696 |
+
"step": 10480
|
| 7697 |
+
},
|
| 7698 |
+
{
|
| 7699 |
+
"epoch": 1.8824281580187092,
|
| 7700 |
+
"grad_norm": 1.1115531921386719,
|
| 7701 |
+
"learning_rate": 6.205137891953155e-06,
|
| 7702 |
+
"loss": 0.5330406188964844,
|
| 7703 |
+
"step": 10490
|
| 7704 |
+
},
|
| 7705 |
+
{
|
| 7706 |
+
"epoch": 1.884222805482648,
|
| 7707 |
+
"grad_norm": 1.319922685623169,
|
| 7708 |
+
"learning_rate": 6.110691348696638e-06,
|
| 7709 |
+
"loss": 0.5549107551574707,
|
| 7710 |
+
"step": 10500
|
| 7711 |
+
},
|
| 7712 |
+
{
|
| 7713 |
+
"epoch": 1.8860174529465867,
|
| 7714 |
+
"grad_norm": 1.9536945819854736,
|
| 7715 |
+
"learning_rate": 6.016244805440121e-06,
|
| 7716 |
+
"loss": 0.5791887760162353,
|
| 7717 |
+
"step": 10510
|
| 7718 |
+
},
|
| 7719 |
+
{
|
| 7720 |
+
"epoch": 1.8878121004105255,
|
| 7721 |
+
"grad_norm": 3.1619863510131836,
|
| 7722 |
+
"learning_rate": 5.921798262183605e-06,
|
| 7723 |
+
"loss": 0.6613367557525635,
|
| 7724 |
+
"step": 10520
|
| 7725 |
+
},
|
| 7726 |
+
{
|
| 7727 |
+
"epoch": 1.8896067478744643,
|
| 7728 |
+
"grad_norm": 0.9547016620635986,
|
| 7729 |
+
"learning_rate": 5.827351718927088e-06,
|
| 7730 |
+
"loss": 0.6350044727325439,
|
| 7731 |
+
"step": 10530
|
| 7732 |
+
},
|
| 7733 |
+
{
|
| 7734 |
+
"epoch": 1.8914013953384032,
|
| 7735 |
+
"grad_norm": 1.2142947912216187,
|
| 7736 |
+
"learning_rate": 5.732905175670571e-06,
|
| 7737 |
+
"loss": 0.5634278297424317,
|
| 7738 |
+
"step": 10540
|
| 7739 |
+
},
|
| 7740 |
+
{
|
| 7741 |
+
"epoch": 1.893196042802342,
|
| 7742 |
+
"grad_norm": 1.3977869749069214,
|
| 7743 |
+
"learning_rate": 5.638458632414054e-06,
|
| 7744 |
+
"loss": 0.5466588497161865,
|
| 7745 |
+
"step": 10550
|
| 7746 |
+
},
|
| 7747 |
+
{
|
| 7748 |
+
"epoch": 1.894990690266281,
|
| 7749 |
+
"grad_norm": 1.9723371267318726,
|
| 7750 |
+
"learning_rate": 5.544012089157537e-06,
|
| 7751 |
+
"loss": 0.582418966293335,
|
| 7752 |
+
"step": 10560
|
| 7753 |
+
},
|
| 7754 |
+
{
|
| 7755 |
+
"epoch": 1.8967853377302197,
|
| 7756 |
+
"grad_norm": 3.2178287506103516,
|
| 7757 |
+
"learning_rate": 5.4495655459010204e-06,
|
| 7758 |
+
"loss": 0.6542830467224121,
|
| 7759 |
+
"step": 10570
|
| 7760 |
+
},
|
| 7761 |
+
{
|
| 7762 |
+
"epoch": 1.8985799851941585,
|
| 7763 |
+
"grad_norm": 0.9720039963722229,
|
| 7764 |
+
"learning_rate": 5.355119002644504e-06,
|
| 7765 |
+
"loss": 0.595330286026001,
|
| 7766 |
+
"step": 10580
|
| 7767 |
+
},
|
| 7768 |
+
{
|
| 7769 |
+
"epoch": 1.9003746326580973,
|
| 7770 |
+
"grad_norm": 1.4674220085144043,
|
| 7771 |
+
"learning_rate": 5.2606724593879864e-06,
|
| 7772 |
+
"loss": 0.552364444732666,
|
| 7773 |
+
"step": 10590
|
| 7774 |
+
},
|
| 7775 |
+
{
|
| 7776 |
+
"epoch": 1.902169280122036,
|
| 7777 |
+
"grad_norm": 1.6370841264724731,
|
| 7778 |
+
"learning_rate": 5.16622591613147e-06,
|
| 7779 |
+
"loss": 0.5085794925689697,
|
| 7780 |
+
"step": 10600
|
| 7781 |
+
},
|
| 7782 |
+
{
|
| 7783 |
+
"epoch": 1.9039639275859748,
|
| 7784 |
+
"grad_norm": 1.807144045829773,
|
| 7785 |
+
"learning_rate": 5.071779372874953e-06,
|
| 7786 |
+
"loss": 0.6325456619262695,
|
| 7787 |
+
"step": 10610
|
| 7788 |
+
},
|
| 7789 |
+
{
|
| 7790 |
+
"epoch": 1.9057585750499135,
|
| 7791 |
+
"grad_norm": 3.3207271099090576,
|
| 7792 |
+
"learning_rate": 4.977332829618436e-06,
|
| 7793 |
+
"loss": 0.7372753620147705,
|
| 7794 |
+
"step": 10620
|
| 7795 |
+
},
|
| 7796 |
+
{
|
| 7797 |
+
"epoch": 1.9075532225138523,
|
| 7798 |
+
"grad_norm": 1.0683008432388306,
|
| 7799 |
+
"learning_rate": 4.882886286361919e-06,
|
| 7800 |
+
"loss": 0.6883942127227783,
|
| 7801 |
+
"step": 10630
|
| 7802 |
+
},
|
| 7803 |
+
{
|
| 7804 |
+
"epoch": 1.9093478699777913,
|
| 7805 |
+
"grad_norm": 1.2829192876815796,
|
| 7806 |
+
"learning_rate": 4.788439743105403e-06,
|
| 7807 |
+
"loss": 0.5925806045532227,
|
| 7808 |
+
"step": 10640
|
| 7809 |
+
},
|
| 7810 |
+
{
|
| 7811 |
+
"epoch": 1.91114251744173,
|
| 7812 |
+
"grad_norm": 1.450143575668335,
|
| 7813 |
+
"learning_rate": 4.693993199848886e-06,
|
| 7814 |
+
"loss": 0.5733682632446289,
|
| 7815 |
+
"step": 10650
|
| 7816 |
+
},
|
| 7817 |
+
{
|
| 7818 |
+
"epoch": 1.912937164905669,
|
| 7819 |
+
"grad_norm": 1.820601224899292,
|
| 7820 |
+
"learning_rate": 4.599546656592369e-06,
|
| 7821 |
+
"loss": 0.58369460105896,
|
| 7822 |
+
"step": 10660
|
| 7823 |
+
},
|
| 7824 |
+
{
|
| 7825 |
+
"epoch": 1.9147318123696078,
|
| 7826 |
+
"grad_norm": 2.5320093631744385,
|
| 7827 |
+
"learning_rate": 4.505100113335852e-06,
|
| 7828 |
+
"loss": 0.679317569732666,
|
| 7829 |
+
"step": 10670
|
| 7830 |
+
},
|
| 7831 |
+
{
|
| 7832 |
+
"epoch": 1.9165264598335465,
|
| 7833 |
+
"grad_norm": 0.9926023483276367,
|
| 7834 |
+
"learning_rate": 4.4106535700793355e-06,
|
| 7835 |
+
"loss": 0.6932086944580078,
|
| 7836 |
+
"step": 10680
|
| 7837 |
+
},
|
| 7838 |
+
{
|
| 7839 |
+
"epoch": 1.9183211072974853,
|
| 7840 |
+
"grad_norm": 1.1948188543319702,
|
| 7841 |
+
"learning_rate": 4.316207026822818e-06,
|
| 7842 |
+
"loss": 0.5830455303192139,
|
| 7843 |
+
"step": 10690
|
| 7844 |
+
},
|
| 7845 |
+
{
|
| 7846 |
+
"epoch": 1.920115754761424,
|
| 7847 |
+
"grad_norm": 1.4618263244628906,
|
| 7848 |
+
"learning_rate": 4.2217604835663015e-06,
|
| 7849 |
+
"loss": 0.5638025283813477,
|
| 7850 |
+
"step": 10700
|
| 7851 |
+
},
|
| 7852 |
+
{
|
| 7853 |
+
"epoch": 1.9219104022253628,
|
| 7854 |
+
"grad_norm": 1.774045705795288,
|
| 7855 |
+
"learning_rate": 4.127313940309785e-06,
|
| 7856 |
+
"loss": 0.6321718215942382,
|
| 7857 |
+
"step": 10710
|
| 7858 |
+
},
|
| 7859 |
+
{
|
| 7860 |
+
"epoch": 1.9237050496893016,
|
| 7861 |
+
"grad_norm": 3.2375879287719727,
|
| 7862 |
+
"learning_rate": 4.0328673970532675e-06,
|
| 7863 |
+
"loss": 0.6737252235412597,
|
| 7864 |
+
"step": 10720
|
| 7865 |
+
},
|
| 7866 |
+
{
|
| 7867 |
+
"epoch": 1.9254996971532403,
|
| 7868 |
+
"grad_norm": 1.0933136940002441,
|
| 7869 |
+
"learning_rate": 3.938420853796751e-06,
|
| 7870 |
+
"loss": 0.6336686134338378,
|
| 7871 |
+
"step": 10730
|
| 7872 |
+
},
|
| 7873 |
+
{
|
| 7874 |
+
"epoch": 1.9272943446171793,
|
| 7875 |
+
"grad_norm": 1.2221369743347168,
|
| 7876 |
+
"learning_rate": 3.843974310540234e-06,
|
| 7877 |
+
"loss": 0.5601771831512451,
|
| 7878 |
+
"step": 10740
|
| 7879 |
+
},
|
| 7880 |
+
{
|
| 7881 |
+
"epoch": 1.929088992081118,
|
| 7882 |
+
"grad_norm": 1.435134768486023,
|
| 7883 |
+
"learning_rate": 3.749527767283718e-06,
|
| 7884 |
+
"loss": 0.518457555770874,
|
| 7885 |
+
"step": 10750
|
| 7886 |
+
},
|
| 7887 |
+
{
|
| 7888 |
+
"epoch": 1.9308836395450568,
|
| 7889 |
+
"grad_norm": 1.5689584016799927,
|
| 7890 |
+
"learning_rate": 3.655081224027201e-06,
|
| 7891 |
+
"loss": 0.5809463500976563,
|
| 7892 |
+
"step": 10760
|
| 7893 |
+
},
|
| 7894 |
+
{
|
| 7895 |
+
"epoch": 1.9326782870089958,
|
| 7896 |
+
"grad_norm": 3.4012184143066406,
|
| 7897 |
+
"learning_rate": 3.560634680770684e-06,
|
| 7898 |
+
"loss": 0.6967726707458496,
|
| 7899 |
+
"step": 10770
|
| 7900 |
+
},
|
| 7901 |
+
{
|
| 7902 |
+
"epoch": 1.9344729344729346,
|
| 7903 |
+
"grad_norm": 1.014663815498352,
|
| 7904 |
+
"learning_rate": 3.4661881375141676e-06,
|
| 7905 |
+
"loss": 0.6451082706451416,
|
| 7906 |
+
"step": 10780
|
| 7907 |
+
},
|
| 7908 |
+
{
|
| 7909 |
+
"epoch": 1.9362675819368733,
|
| 7910 |
+
"grad_norm": 1.2918916940689087,
|
| 7911 |
+
"learning_rate": 3.3717415942576506e-06,
|
| 7912 |
+
"loss": 0.5296636104583741,
|
| 7913 |
+
"step": 10790
|
| 7914 |
+
},
|
| 7915 |
+
{
|
| 7916 |
+
"epoch": 1.938062229400812,
|
| 7917 |
+
"grad_norm": 1.5131940841674805,
|
| 7918 |
+
"learning_rate": 3.2772950510011336e-06,
|
| 7919 |
+
"loss": 0.5459217071533203,
|
| 7920 |
+
"step": 10800
|
| 7921 |
+
},
|
| 7922 |
+
{
|
| 7923 |
+
"epoch": 1.9398568768647508,
|
| 7924 |
+
"grad_norm": 1.8521559238433838,
|
| 7925 |
+
"learning_rate": 3.1828485077446166e-06,
|
| 7926 |
+
"loss": 0.6091119289398194,
|
| 7927 |
+
"step": 10810
|
| 7928 |
+
},
|
| 7929 |
+
{
|
| 7930 |
+
"epoch": 1.9416515243286896,
|
| 7931 |
+
"grad_norm": 3.2888715267181396,
|
| 7932 |
+
"learning_rate": 3.0884019644881e-06,
|
| 7933 |
+
"loss": 0.6569681167602539,
|
| 7934 |
+
"step": 10820
|
| 7935 |
+
},
|
| 7936 |
+
{
|
| 7937 |
+
"epoch": 1.9434461717926284,
|
| 7938 |
+
"grad_norm": 0.9157596826553345,
|
| 7939 |
+
"learning_rate": 2.993955421231583e-06,
|
| 7940 |
+
"loss": 0.6697529315948486,
|
| 7941 |
+
"step": 10830
|
| 7942 |
+
},
|
| 7943 |
+
{
|
| 7944 |
+
"epoch": 1.9452408192565673,
|
| 7945 |
+
"grad_norm": 1.2682820558547974,
|
| 7946 |
+
"learning_rate": 2.8995088779750665e-06,
|
| 7947 |
+
"loss": 0.55664963722229,
|
| 7948 |
+
"step": 10840
|
| 7949 |
+
},
|
| 7950 |
+
{
|
| 7951 |
+
"epoch": 1.947035466720506,
|
| 7952 |
+
"grad_norm": 1.4419405460357666,
|
| 7953 |
+
"learning_rate": 2.8050623347185495e-06,
|
| 7954 |
+
"loss": 0.5564156532287597,
|
| 7955 |
+
"step": 10850
|
| 7956 |
+
},
|
| 7957 |
+
{
|
| 7958 |
+
"epoch": 1.9488301141844449,
|
| 7959 |
+
"grad_norm": 1.7976231575012207,
|
| 7960 |
+
"learning_rate": 2.710615791462033e-06,
|
| 7961 |
+
"loss": 0.5685441493988037,
|
| 7962 |
+
"step": 10860
|
| 7963 |
+
},
|
| 7964 |
+
{
|
| 7965 |
+
"epoch": 1.9506247616483838,
|
| 7966 |
+
"grad_norm": 3.030071496963501,
|
| 7967 |
+
"learning_rate": 2.616169248205516e-06,
|
| 7968 |
+
"loss": 0.7275540828704834,
|
| 7969 |
+
"step": 10870
|
| 7970 |
+
},
|
| 7971 |
+
{
|
| 7972 |
+
"epoch": 1.9524194091123226,
|
| 7973 |
+
"grad_norm": 0.9376201629638672,
|
| 7974 |
+
"learning_rate": 2.5217227049489993e-06,
|
| 7975 |
+
"loss": 0.5742666244506835,
|
| 7976 |
+
"step": 10880
|
| 7977 |
+
},
|
| 7978 |
+
{
|
| 7979 |
+
"epoch": 1.9542140565762613,
|
| 7980 |
+
"grad_norm": 1.2221745252609253,
|
| 7981 |
+
"learning_rate": 2.4272761616924823e-06,
|
| 7982 |
+
"loss": 0.5613106250762939,
|
| 7983 |
+
"step": 10890
|
| 7984 |
+
},
|
| 7985 |
+
{
|
| 7986 |
+
"epoch": 1.9560087040402,
|
| 7987 |
+
"grad_norm": 1.2180469036102295,
|
| 7988 |
+
"learning_rate": 2.3328296184359653e-06,
|
| 7989 |
+
"loss": 0.5035813808441162,
|
| 7990 |
+
"step": 10900
|
| 7991 |
+
},
|
| 7992 |
+
{
|
| 7993 |
+
"epoch": 1.9578033515041389,
|
| 7994 |
+
"grad_norm": 1.8794699907302856,
|
| 7995 |
+
"learning_rate": 2.2383830751794487e-06,
|
| 7996 |
+
"loss": 0.6061521053314209,
|
| 7997 |
+
"step": 10910
|
| 7998 |
+
},
|
| 7999 |
+
{
|
| 8000 |
+
"epoch": 1.9595979989680776,
|
| 8001 |
+
"grad_norm": 2.521310329437256,
|
| 8002 |
+
"learning_rate": 2.1439365319229317e-06,
|
| 8003 |
+
"loss": 0.6348999977111817,
|
| 8004 |
+
"step": 10920
|
| 8005 |
+
},
|
| 8006 |
+
{
|
| 8007 |
+
"epoch": 1.9613926464320164,
|
| 8008 |
+
"grad_norm": 0.9879629611968994,
|
| 8009 |
+
"learning_rate": 2.049489988666415e-06,
|
| 8010 |
+
"loss": 0.6965922355651856,
|
| 8011 |
+
"step": 10930
|
| 8012 |
+
},
|
| 8013 |
+
{
|
| 8014 |
+
"epoch": 1.9631872938959554,
|
| 8015 |
+
"grad_norm": 1.1919386386871338,
|
| 8016 |
+
"learning_rate": 1.955043445409898e-06,
|
| 8017 |
+
"loss": 0.5606507778167724,
|
| 8018 |
+
"step": 10940
|
| 8019 |
+
},
|
| 8020 |
+
{
|
| 8021 |
+
"epoch": 1.9649819413598941,
|
| 8022 |
+
"grad_norm": 1.341180682182312,
|
| 8023 |
+
"learning_rate": 1.8605969021533814e-06,
|
| 8024 |
+
"loss": 0.519660758972168,
|
| 8025 |
+
"step": 10950
|
| 8026 |
+
},
|
| 8027 |
+
{
|
| 8028 |
+
"epoch": 1.9667765888238329,
|
| 8029 |
+
"grad_norm": 1.7559319734573364,
|
| 8030 |
+
"learning_rate": 1.7661503588968644e-06,
|
| 8031 |
+
"loss": 0.6203202724456787,
|
| 8032 |
+
"step": 10960
|
| 8033 |
+
},
|
| 8034 |
+
{
|
| 8035 |
+
"epoch": 1.9685712362877719,
|
| 8036 |
+
"grad_norm": 2.916363477706909,
|
| 8037 |
+
"learning_rate": 1.6717038156403478e-06,
|
| 8038 |
+
"loss": 0.667643404006958,
|
| 8039 |
+
"step": 10970
|
| 8040 |
+
},
|
| 8041 |
+
{
|
| 8042 |
+
"epoch": 1.9703658837517106,
|
| 8043 |
+
"grad_norm": 0.9973469972610474,
|
| 8044 |
+
"learning_rate": 1.5772572723838308e-06,
|
| 8045 |
+
"loss": 0.6697774410247803,
|
| 8046 |
+
"step": 10980
|
| 8047 |
+
},
|
| 8048 |
+
{
|
| 8049 |
+
"epoch": 1.9721605312156494,
|
| 8050 |
+
"grad_norm": 1.3039273023605347,
|
| 8051 |
+
"learning_rate": 1.482810729127314e-06,
|
| 8052 |
+
"loss": 0.5913773536682129,
|
| 8053 |
+
"step": 10990
|
| 8054 |
+
},
|
| 8055 |
+
{
|
| 8056 |
+
"epoch": 1.9739551786795881,
|
| 8057 |
+
"grad_norm": 1.738749384880066,
|
| 8058 |
+
"learning_rate": 1.3883641858707972e-06,
|
| 8059 |
+
"loss": 0.589254903793335,
|
| 8060 |
+
"step": 11000
|
| 8061 |
}
|
| 8062 |
],
|
| 8063 |
"logging_steps": 10,
|
|
|
|
| 8077 |
"attributes": {}
|
| 8078 |
}
|
| 8079 |
},
|
| 8080 |
+
"total_flos": 1.1205709942418235e+18,
|
| 8081 |
"train_batch_size": 2,
|
| 8082 |
"trial_name": null,
|
| 8083 |
"trial_params": null
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-8000 → checkpoint-11000}/training_args.bin
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/tb/events.out.tfevents.1790426135.b0aacb224392.4711.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aaa5b53fa51a721eec266ac109a785495654b087e7343f2a41dd8ec89968329f
|
| 3 |
+
size 257783
|
qwen3.5-2b-50pct-masked-weighted/test_eval_entity/metrics.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"wer": 0.
|
| 3 |
-
"wer_zwnj_normalized": 0.
|
| 4 |
"exact_match": 0.0029702970297029703,
|
| 5 |
-
"fix_rate": 0.
|
| 6 |
-
"preservation_rate": 0.
|
| 7 |
-
"targeted_score": 0.
|
| 8 |
-
"fix_rate_lenient": 0.
|
| 9 |
-
"preservation_rate_lenient": 0.
|
| 10 |
-
"targeted_score_lenient": 0.
|
| 11 |
-
"fix_rate_recoverable": 0.
|
| 12 |
"unrecoverable_targets": 523,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
"n_examples": 1010
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.33916696021605236,
|
| 3 |
+
"wer_zwnj_normalized": 0.3351746885707881,
|
| 4 |
"exact_match": 0.0029702970297029703,
|
| 5 |
+
"fix_rate": 0.27269311456930834,
|
| 6 |
+
"preservation_rate": 0.7685507479491717,
|
| 7 |
+
"targeted_score": 0.52062193125924,
|
| 8 |
+
"fix_rate_lenient": 0.26436781609195403,
|
| 9 |
+
"preservation_rate_lenient": 0.7709672452946383,
|
| 10 |
+
"targeted_score_lenient": 0.5176675306932962,
|
| 11 |
+
"fix_rate_recoverable": 0.2682193337120326,
|
| 12 |
"unrecoverable_targets": 523,
|
| 13 |
+
"hallucination_rate": 0.0049504950495049506,
|
| 14 |
"n_examples": 1010
|
| 15 |
}
|
qwen3.5-2b-50pct-masked-weighted/test_eval_entity/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
qwen3.5-2b-50pct-masked-weighted/test_eval_typo/metrics.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"wer": 0.
|
| 3 |
-
"wer_zwnj_normalized": 0.
|
| 4 |
-
"exact_match": 0.
|
| 5 |
-
"fix_rate": 0.
|
| 6 |
-
"preservation_rate": 0.
|
| 7 |
-
"targeted_score": 0.
|
| 8 |
-
"fix_rate_lenient": 0.
|
| 9 |
-
"preservation_rate_lenient": 0.
|
| 10 |
-
"targeted_score_lenient": 0.
|
| 11 |
-
"fix_rate_recoverable": 0.
|
| 12 |
"unrecoverable_targets": 73,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
"n_examples": 358
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.22427305915090245,
|
| 3 |
+
"wer_zwnj_normalized": 0.21903959561920808,
|
| 4 |
+
"exact_match": 0.0111731843575419,
|
| 5 |
+
"fix_rate": 0.369716944552152,
|
| 6 |
+
"preservation_rate": 0.8756357863169965,
|
| 7 |
+
"targeted_score": 0.6226763654345742,
|
| 8 |
+
"fix_rate_lenient": 0.3618298488925688,
|
| 9 |
+
"preservation_rate_lenient": 0.8781095256233702,
|
| 10 |
+
"targeted_score_lenient": 0.6199696872579695,
|
| 11 |
+
"fix_rate_recoverable": 0.366120218579235,
|
| 12 |
"unrecoverable_targets": 73,
|
| 13 |
+
"hallucination_rate": 0.013966480446927373,
|
| 14 |
"n_examples": 358
|
| 15 |
}
|
qwen3.5-2b-50pct-masked-weighted/test_eval_typo/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|