Upload folder using huggingface_hub
Browse files- checkpoints-v6.0-discrete/checkpoint-21504/model.safetensors +3 -0
- checkpoints-v6.0-discrete/checkpoint-21504/optimizer.pt +3 -0
- checkpoints-v6.0-discrete/checkpoint-21504/rng_state.pth +3 -0
- checkpoints-v6.0-discrete/checkpoint-21504/scaler.pt +3 -0
- checkpoints-v6.0-discrete/checkpoint-21504/scheduler.pt +3 -0
- checkpoints-v6.0-discrete/checkpoint-21504/trainer_state.json +979 -0
- checkpoints-v6.0-discrete/checkpoint-21504/training_args.bin +3 -0
checkpoints-v6.0-discrete/checkpoint-21504/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ed101bca614841c16e2f8d32a3a342d09a01118c84e154c113ed75ef1916444
|
| 3 |
+
size 24033312
|
checkpoints-v6.0-discrete/checkpoint-21504/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:82b2b7371eceac24a1831bcea6aaf4951e027ac3d1d61b1aae3b2bbe977646e0
|
| 3 |
+
size 1641995
|
checkpoints-v6.0-discrete/checkpoint-21504/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7dc4ccae59b934bd4864e6262348ab34d1c9526f18f0af9afda6107b47b8fc48
|
| 3 |
+
size 14645
|
checkpoints-v6.0-discrete/checkpoint-21504/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13ee8ccd0a21afad67dde5fcddb164532e45a943c29e0004f1ac286f19b616ef
|
| 3 |
+
size 1383
|
checkpoints-v6.0-discrete/checkpoint-21504/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c8d0f20f63f45b1b1775e9f97d9c6a1a2904fa04b7d5a41f30f11b9aef0479eb
|
| 3 |
+
size 1465
|
checkpoints-v6.0-discrete/checkpoint-21504/trainer_state.json
ADDED
|
@@ -0,0 +1,979 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.9932104752667313,
|
| 6 |
+
"eval_steps": 1024,
|
| 7 |
+
"global_step": 21504,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.011823934229365849,
|
| 14 |
+
"grad_norm": 1.512515902519226,
|
| 15 |
+
"learning_rate": 0.000498046875,
|
| 16 |
+
"loss": 0.8726199269294739,
|
| 17 |
+
"step": 256
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.023647868458731697,
|
| 21 |
+
"grad_norm": 3.4460880756378174,
|
| 22 |
+
"learning_rate": 0.000998046875,
|
| 23 |
+
"loss": 0.256637305021286,
|
| 24 |
+
"step": 512
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.03547180268809755,
|
| 28 |
+
"grad_norm": 1.2741448879241943,
|
| 29 |
+
"learning_rate": 0.000999640996023194,
|
| 30 |
+
"loss": 0.13637158274650574,
|
| 31 |
+
"step": 768
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.047295736917463395,
|
| 35 |
+
"grad_norm": 0.3841530680656433,
|
| 36 |
+
"learning_rate": 0.0009985588674043958,
|
| 37 |
+
"loss": 0.08800904452800751,
|
| 38 |
+
"step": 1024
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.047295736917463395,
|
| 42 |
+
"eval_cos_loss": 0.034078707212590736,
|
| 43 |
+
"eval_loss": 0.07364735873968906,
|
| 44 |
+
"eval_mse_loss": 0.07023948820554503,
|
| 45 |
+
"step": 1024
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.047295736917463395,
|
| 49 |
+
"eval_cos_loss": 0.034078707212590736,
|
| 50 |
+
"eval_loss": 0.07364735873968906,
|
| 51 |
+
"eval_mse_loss": 0.07023948820554503,
|
| 52 |
+
"eval_runtime": 48.1981,
|
| 53 |
+
"eval_samples_per_second": 580.791,
|
| 54 |
+
"eval_steps_per_second": 9.088,
|
| 55 |
+
"step": 1024
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.05911967114682925,
|
| 59 |
+
"grad_norm": 0.9900110363960266,
|
| 60 |
+
"learning_rate": 0.0009967551747861387,
|
| 61 |
+
"loss": 0.0682656541466713,
|
| 62 |
+
"step": 1280
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.0709436053761951,
|
| 66 |
+
"grad_norm": 1.70609450340271,
|
| 67 |
+
"learning_rate": 0.000994232528651847,
|
| 68 |
+
"loss": 0.053092263638973236,
|
| 69 |
+
"step": 1536
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08276753960556095,
|
| 73 |
+
"grad_norm": 0.5201518535614014,
|
| 74 |
+
"learning_rate": 0.0009909945800260092,
|
| 75 |
+
"loss": 0.04387389123439789,
|
| 76 |
+
"step": 1792
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.09459147383492679,
|
| 80 |
+
"grad_norm": 0.8182135224342346,
|
| 81 |
+
"learning_rate": 0.0009870460151900522,
|
| 82 |
+
"loss": 0.03948979824781418,
|
| 83 |
+
"step": 2048
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.09459147383492679,
|
| 87 |
+
"eval_cos_loss": 0.016229083018428455,
|
| 88 |
+
"eval_loss": 0.03677510998751883,
|
| 89 |
+
"eval_mse_loss": 0.035152201642830894,
|
| 90 |
+
"step": 2048
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09459147383492679,
|
| 94 |
+
"eval_cos_loss": 0.016229083018428455,
|
| 95 |
+
"eval_loss": 0.03677510998751883,
|
| 96 |
+
"eval_mse_loss": 0.035152201642830894,
|
| 97 |
+
"eval_runtime": 48.5188,
|
| 98 |
+
"eval_samples_per_second": 576.952,
|
| 99 |
+
"eval_steps_per_second": 9.027,
|
| 100 |
+
"step": 2048
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.10641540806429264,
|
| 104 |
+
"grad_norm": 0.32195019721984863,
|
| 105 |
+
"learning_rate": 0.0009823925488998885,
|
| 106 |
+
"loss": 0.03621590510010719,
|
| 107 |
+
"step": 2304
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.1182393422936585,
|
| 111 |
+
"grad_norm": 0.30791640281677246,
|
| 112 |
+
"learning_rate": 0.0009770409161149525,
|
| 113 |
+
"loss": 0.03272327035665512,
|
| 114 |
+
"step": 2560
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.13006327652302435,
|
| 118 |
+
"grad_norm": 2.691577434539795,
|
| 119 |
+
"learning_rate": 0.0009709988622506973,
|
| 120 |
+
"loss": 0.029315859079360962,
|
| 121 |
+
"step": 2816
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.1418872107523902,
|
| 125 |
+
"grad_norm": 0.7709335088729858,
|
| 126 |
+
"learning_rate": 0.000964275131968659,
|
| 127 |
+
"loss": 0.026046201586723328,
|
| 128 |
+
"step": 3072
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.1418872107523902,
|
| 132 |
+
"eval_cos_loss": 0.010817144368562558,
|
| 133 |
+
"eval_loss": 0.02589840943112833,
|
| 134 |
+
"eval_mse_loss": 0.024816695003255725,
|
| 135 |
+
"step": 3072
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1418872107523902,
|
| 139 |
+
"eval_cos_loss": 0.010817144368562558,
|
| 140 |
+
"eval_loss": 0.02589840943112833,
|
| 141 |
+
"eval_mse_loss": 0.024816695003255725,
|
| 142 |
+
"eval_runtime": 48.4081,
|
| 143 |
+
"eval_samples_per_second": 578.271,
|
| 144 |
+
"eval_steps_per_second": 9.048,
|
| 145 |
+
"step": 3072
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.15371114498175603,
|
| 149 |
+
"grad_norm": 0.7047502994537354,
|
| 150 |
+
"learning_rate": 0.0009568794565203123,
|
| 151 |
+
"loss": 0.02368602715432644,
|
| 152 |
+
"step": 3328
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.1655350792111219,
|
| 156 |
+
"grad_norm": 0.28145521879196167,
|
| 157 |
+
"learning_rate": 0.0009488225396630347,
|
| 158 |
+
"loss": 0.02464677020907402,
|
| 159 |
+
"step": 3584
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.17735901344048774,
|
| 163 |
+
"grad_norm": 0.5821279883384705,
|
| 164 |
+
"learning_rate": 0.0009401160421685646,
|
| 165 |
+
"loss": 0.02320299856364727,
|
| 166 |
+
"step": 3840
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.18918294766985358,
|
| 170 |
+
"grad_norm": 0.254409521818161,
|
| 171 |
+
"learning_rate": 0.0009307725649463714,
|
| 172 |
+
"loss": 0.02275155484676361,
|
| 173 |
+
"step": 4096
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.18918294766985358,
|
| 177 |
+
"eval_cos_loss": 0.009839889010548438,
|
| 178 |
+
"eval_loss": 0.023258781913225644,
|
| 179 |
+
"eval_mse_loss": 0.02227479305552089,
|
| 180 |
+
"step": 4096
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.18918294766985358,
|
| 184 |
+
"eval_cos_loss": 0.009839889010548438,
|
| 185 |
+
"eval_loss": 0.023258781913225644,
|
| 186 |
+
"eval_mse_loss": 0.02227479305552089,
|
| 187 |
+
"eval_runtime": 48.1848,
|
| 188 |
+
"eval_samples_per_second": 580.951,
|
| 189 |
+
"eval_steps_per_second": 9.09,
|
| 190 |
+
"step": 4096
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.20100688189921945,
|
| 194 |
+
"grad_norm": 0.23914200067520142,
|
| 195 |
+
"learning_rate": 0.0009208056308063659,
|
| 196 |
+
"loss": 0.021996984258294106,
|
| 197 |
+
"step": 4352
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.2128308161285853,
|
| 201 |
+
"grad_norm": 0.3439199924468994,
|
| 202 |
+
"learning_rate": 0.0009102296648873445,
|
| 203 |
+
"loss": 0.022452646866440773,
|
| 204 |
+
"step": 4608
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.22465475035795113,
|
| 208 |
+
"grad_norm": 0.3184088468551636,
|
| 209 |
+
"learning_rate": 0.0008990599737794927,
|
| 210 |
+
"loss": 0.021988332271575928,
|
| 211 |
+
"step": 4864
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.236478684587317,
|
| 215 |
+
"grad_norm": 0.19294844567775726,
|
| 216 |
+
"learning_rate": 0.0008873127233711644,
|
| 217 |
+
"loss": 0.021919861435890198,
|
| 218 |
+
"step": 5120
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.236478684587317,
|
| 222 |
+
"eval_cos_loss": 0.008801625348621188,
|
| 223 |
+
"eval_loss": 0.020690152113279115,
|
| 224 |
+
"eval_mse_loss": 0.019809989538256367,
|
| 225 |
+
"step": 5120
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.236478684587317,
|
| 229 |
+
"eval_cos_loss": 0.008801625348621188,
|
| 230 |
+
"eval_loss": 0.020690152113279115,
|
| 231 |
+
"eval_mse_loss": 0.019809989538256367,
|
| 232 |
+
"eval_runtime": 48.3851,
|
| 233 |
+
"eval_samples_per_second": 578.546,
|
| 234 |
+
"eval_steps_per_second": 9.052,
|
| 235 |
+
"step": 5120
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.24830261881668284,
|
| 239 |
+
"grad_norm": 0.42197832465171814,
|
| 240 |
+
"learning_rate": 0.0008750049154520011,
|
| 241 |
+
"loss": 0.02057112567126751,
|
| 242 |
+
"step": 5376
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.2601265530460487,
|
| 246 |
+
"grad_norm": 0.5257790684700012,
|
| 247 |
+
"learning_rate": 0.0008621543631062487,
|
| 248 |
+
"loss": 0.020898405462503433,
|
| 249 |
+
"step": 5632
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.27195048727541454,
|
| 253 |
+
"grad_norm": 0.26698386669158936,
|
| 254 |
+
"learning_rate": 0.0008487796649318904,
|
| 255 |
+
"loss": 0.020850826054811478,
|
| 256 |
+
"step": 5888
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.2837744215047804,
|
| 260 |
+
"grad_norm": 0.08679121732711792,
|
| 261 |
+
"learning_rate": 0.0008349001781229053,
|
| 262 |
+
"loss": 0.020534055307507515,
|
| 263 |
+
"step": 6144
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.2837744215047804,
|
| 267 |
+
"eval_cos_loss": 0.008999818267532492,
|
| 268 |
+
"eval_loss": 0.02102939254438428,
|
| 269 |
+
"eval_mse_loss": 0.020129410714787094,
|
| 270 |
+
"step": 6144
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.2837744215047804,
|
| 274 |
+
"eval_cos_loss": 0.008999818267532492,
|
| 275 |
+
"eval_loss": 0.02102939254438428,
|
| 276 |
+
"eval_mse_loss": 0.020129410714787094,
|
| 277 |
+
"eval_runtime": 48.2922,
|
| 278 |
+
"eval_samples_per_second": 579.659,
|
| 279 |
+
"eval_steps_per_second": 9.07,
|
| 280 |
+
"step": 6144
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.2955983557341462,
|
| 284 |
+
"grad_norm": 0.19946622848510742,
|
| 285 |
+
"learning_rate": 0.0008205359904536107,
|
| 286 |
+
"loss": 0.021952064707875252,
|
| 287 |
+
"step": 6400
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.30742228996351206,
|
| 291 |
+
"grad_norm": 0.3316427171230316,
|
| 292 |
+
"learning_rate": 0.0008057078912056363,
|
| 293 |
+
"loss": 0.020055362954735756,
|
| 294 |
+
"step": 6656
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.3192462241928779,
|
| 298 |
+
"grad_norm": 0.1748799830675125,
|
| 299 |
+
"learning_rate": 0.0007904373410796086,
|
| 300 |
+
"loss": 0.01934388093650341,
|
| 301 |
+
"step": 6912
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.3310701584222438,
|
| 305 |
+
"grad_norm": 0.2375132441520691,
|
| 306 |
+
"learning_rate": 0.0007747464411350876,
|
| 307 |
+
"loss": 0.01997610181570053,
|
| 308 |
+
"step": 7168
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.3310701584222438,
|
| 312 |
+
"eval_cos_loss": 0.008433358508298284,
|
| 313 |
+
"eval_loss": 0.01962870682300532,
|
| 314 |
+
"eval_mse_loss": 0.018785371012415855,
|
| 315 |
+
"step": 7168
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.3310701584222438,
|
| 319 |
+
"eval_cos_loss": 0.008433358508298284,
|
| 320 |
+
"eval_loss": 0.01962870682300532,
|
| 321 |
+
"eval_mse_loss": 0.018785371012415855,
|
| 322 |
+
"eval_runtime": 48.9101,
|
| 323 |
+
"eval_samples_per_second": 572.335,
|
| 324 |
+
"eval_steps_per_second": 8.955,
|
| 325 |
+
"step": 7168
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.34289409265160964,
|
| 329 |
+
"grad_norm": 0.2102879285812378,
|
| 330 |
+
"learning_rate": 0.000758657900803716,
|
| 331 |
+
"loss": 0.01965208910405636,
|
| 332 |
+
"step": 7424
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.3547180268809755,
|
| 336 |
+
"grad_norm": 0.15301695466041565,
|
| 337 |
+
"learning_rate": 0.000742195005021869,
|
| 338 |
+
"loss": 0.01973935030400753,
|
| 339 |
+
"step": 7680
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.3665419611103413,
|
| 343 |
+
"grad_norm": 0.15693935751914978,
|
| 344 |
+
"learning_rate": 0.0007253815805303786,
|
| 345 |
+
"loss": 0.019944220781326294,
|
| 346 |
+
"step": 7936
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.37836589533970716,
|
| 350 |
+
"grad_norm": 0.11684704571962357,
|
| 351 |
+
"learning_rate": 0.0007082419613901028,
|
| 352 |
+
"loss": 0.01952940598130226,
|
| 353 |
+
"step": 8192
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.37836589533970716,
|
| 357 |
+
"eval_cos_loss": 0.007886844790821308,
|
| 358 |
+
"eval_loss": 0.018350816901654143,
|
| 359 |
+
"eval_mse_loss": 0.017562132414864148,
|
| 360 |
+
"step": 8192
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.37836589533970716,
|
| 364 |
+
"eval_cos_loss": 0.007886844790821308,
|
| 365 |
+
"eval_loss": 0.018350816901654143,
|
| 366 |
+
"eval_mse_loss": 0.017562132414864148,
|
| 367 |
+
"eval_runtime": 48.7465,
|
| 368 |
+
"eval_samples_per_second": 574.256,
|
| 369 |
+
"eval_steps_per_second": 8.985,
|
| 370 |
+
"step": 8192
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.390189829569073,
|
| 374 |
+
"grad_norm": 0.18092115223407745,
|
| 375 |
+
"learning_rate": 0.0006908009537632514,
|
| 376 |
+
"loss": 0.018511567264795303,
|
| 377 |
+
"step": 8448
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.4020137637984389,
|
| 381 |
+
"grad_norm": 0.35747045278549194,
|
| 382 |
+
"learning_rate": 0.0006730838000114403,
|
| 383 |
+
"loss": 0.018113775178790092,
|
| 384 |
+
"step": 8704
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.41383769802780473,
|
| 388 |
+
"grad_norm": 0.2213900238275528,
|
| 389 |
+
"learning_rate": 0.0006551161421624341,
|
| 390 |
+
"loss": 0.018880348652601242,
|
| 391 |
+
"step": 8960
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.4256616322571706,
|
| 395 |
+
"grad_norm": 0.10871223360300064,
|
| 396 |
+
"learning_rate": 0.0006369239847984517,
|
| 397 |
+
"loss": 0.018961625173687935,
|
| 398 |
+
"step": 9216
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.4256616322571706,
|
| 402 |
+
"eval_cos_loss": 0.007983019030344625,
|
| 403 |
+
"eval_loss": 0.01807575462254181,
|
| 404 |
+
"eval_mse_loss": 0.017277452761953756,
|
| 405 |
+
"step": 9216
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.4256616322571706,
|
| 409 |
+
"eval_cos_loss": 0.007983019030344625,
|
| 410 |
+
"eval_loss": 0.01807575462254181,
|
| 411 |
+
"eval_mse_loss": 0.017277452761953756,
|
| 412 |
+
"eval_runtime": 48.8313,
|
| 413 |
+
"eval_samples_per_second": 573.259,
|
| 414 |
+
"eval_steps_per_second": 8.97,
|
| 415 |
+
"step": 9216
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.4374855664865364,
|
| 419 |
+
"grad_norm": 0.25837093591690063,
|
| 420 |
+
"learning_rate": 0.0006185336574197479,
|
| 421 |
+
"loss": 0.01820954494178295,
|
| 422 |
+
"step": 9472
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.44930950071590225,
|
| 426 |
+
"grad_norm": 0.17237816751003265,
|
| 427 |
+
"learning_rate": 0.0005999717763379407,
|
| 428 |
+
"loss": 0.018442915752530098,
|
| 429 |
+
"step": 9728
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.4611334349452681,
|
| 433 |
+
"grad_norm": 0.3920765817165375,
|
| 434 |
+
"learning_rate": 0.0005812652061542363,
|
| 435 |
+
"loss": 0.018666906282305717,
|
| 436 |
+
"step": 9984
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.472957369174634,
|
| 440 |
+
"grad_norm": 0.13486474752426147,
|
| 441 |
+
"learning_rate": 0.0005624410208783071,
|
| 442 |
+
"loss": 0.018142327666282654,
|
| 443 |
+
"step": 10240
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.472957369174634,
|
| 447 |
+
"eval_cos_loss": 0.007294310057763282,
|
| 448 |
+
"eval_loss": 0.017106381499051162,
|
| 449 |
+
"eval_mse_loss": 0.016376950456994718,
|
| 450 |
+
"step": 10240
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.472957369174634,
|
| 454 |
+
"eval_cos_loss": 0.007294310057763282,
|
| 455 |
+
"eval_loss": 0.017106381499051162,
|
| 456 |
+
"eval_mse_loss": 0.016376950456994718,
|
| 457 |
+
"eval_runtime": 48.2189,
|
| 458 |
+
"eval_samples_per_second": 580.54,
|
| 459 |
+
"eval_steps_per_second": 9.084,
|
| 460 |
+
"step": 10240
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.48478130340399983,
|
| 464 |
+
"grad_norm": 0.12932348251342773,
|
| 465 |
+
"learning_rate": 0.0005435264647440881,
|
| 466 |
+
"loss": 0.01894083432853222,
|
| 467 |
+
"step": 10496
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.49660523763336567,
|
| 471 |
+
"grad_norm": 0.23910421133041382,
|
| 472 |
+
"learning_rate": 0.000524548912779213,
|
| 473 |
+
"loss": 0.01813199371099472,
|
| 474 |
+
"step": 10752
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.5084291718627315,
|
| 478 |
+
"grad_norm": 0.13863646984100342,
|
| 479 |
+
"learning_rate": 0.0005055358311851499,
|
| 480 |
+
"loss": 0.0171187836676836,
|
| 481 |
+
"step": 11008
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 0.5202531060920974,
|
| 485 |
+
"grad_norm": 0.10762029141187668,
|
| 486 |
+
"learning_rate": 0.0004865147375853812,
|
| 487 |
+
"loss": 0.018286000937223434,
|
| 488 |
+
"step": 11264
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.5202531060920974,
|
| 492 |
+
"eval_cos_loss": 0.008126283663040086,
|
| 493 |
+
"eval_loss": 0.018038252006033987,
|
| 494 |
+
"eval_mse_loss": 0.017225623636859463,
|
| 495 |
+
"step": 11264
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.5202531060920974,
|
| 499 |
+
"eval_cos_loss": 0.008126283663040086,
|
| 500 |
+
"eval_loss": 0.018038252006033987,
|
| 501 |
+
"eval_mse_loss": 0.017225623636859463,
|
| 502 |
+
"eval_runtime": 48.1833,
|
| 503 |
+
"eval_samples_per_second": 580.969,
|
| 504 |
+
"eval_steps_per_second": 9.09,
|
| 505 |
+
"step": 11264
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.5320770403214632,
|
| 509 |
+
"grad_norm": 0.1320306956768036,
|
| 510 |
+
"learning_rate": 0.0004675131611991607,
|
| 511 |
+
"loss": 0.017766807228326797,
|
| 512 |
+
"step": 11520
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.5439009745508291,
|
| 516 |
+
"grad_norm": 0.4067651629447937,
|
| 517 |
+
"learning_rate": 0.0004485586029984899,
|
| 518 |
+
"loss": 0.016927646473050117,
|
| 519 |
+
"step": 11776
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.5557249087801949,
|
| 523 |
+
"grad_norm": 0.14531229436397552,
|
| 524 |
+
"learning_rate": 0.00042967849590597266,
|
| 525 |
+
"loss": 0.017305444926023483,
|
| 526 |
+
"step": 12032
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.5675488430095608,
|
| 530 |
+
"grad_norm": 0.10750454664230347,
|
| 531 |
+
"learning_rate": 0.0004109001650911621,
|
| 532 |
+
"loss": 0.017024781554937363,
|
| 533 |
+
"step": 12288
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.5675488430095608,
|
| 537 |
+
"eval_cos_loss": 0.0077252416147480475,
|
| 538 |
+
"eval_loss": 0.017118561693964755,
|
| 539 |
+
"eval_mse_loss": 0.01634603754625779,
|
| 540 |
+
"step": 12288
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.5675488430095608,
|
| 544 |
+
"eval_cos_loss": 0.0077252416147480475,
|
| 545 |
+
"eval_loss": 0.017118561693964755,
|
| 546 |
+
"eval_mse_loss": 0.01634603754625779,
|
| 547 |
+
"eval_runtime": 48.0617,
|
| 548 |
+
"eval_samples_per_second": 582.439,
|
| 549 |
+
"eval_steps_per_second": 9.113,
|
| 550 |
+
"step": 12288
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.5793727772389267,
|
| 554 |
+
"grad_norm": 0.18257588148117065,
|
| 555 |
+
"learning_rate": 0.0003922507884228551,
|
| 556 |
+
"loss": 0.01718640699982643,
|
| 557 |
+
"step": 12544
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.5911967114682924,
|
| 561 |
+
"grad_norm": 0.11715080589056015,
|
| 562 |
+
"learning_rate": 0.00037375735713457723,
|
| 563 |
+
"loss": 0.016841745004057884,
|
| 564 |
+
"step": 12800
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.6030206456976583,
|
| 568 |
+
"grad_norm": 0.14302287995815277,
|
| 569 |
+
"learning_rate": 0.00035544663676018276,
|
| 570 |
+
"loss": 0.016849281266331673,
|
| 571 |
+
"step": 13056
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.6148445799270241,
|
| 575 |
+
"grad_norm": 0.309292733669281,
|
| 576 |
+
"learning_rate": 0.00033734512839611255,
|
| 577 |
+
"loss": 0.01751680113375187,
|
| 578 |
+
"step": 13312
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.6148445799270241,
|
| 582 |
+
"eval_cos_loss": 0.008235950203671983,
|
| 583 |
+
"eval_loss": 0.017767599055577562,
|
| 584 |
+
"eval_mse_loss": 0.016944004034572472,
|
| 585 |
+
"step": 13312
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.6148445799270241,
|
| 589 |
+
"eval_cos_loss": 0.008235950203671983,
|
| 590 |
+
"eval_loss": 0.017767599055577562,
|
| 591 |
+
"eval_mse_loss": 0.016944004034572472,
|
| 592 |
+
"eval_runtime": 48.8465,
|
| 593 |
+
"eval_samples_per_second": 573.081,
|
| 594 |
+
"eval_steps_per_second": 8.967,
|
| 595 |
+
"step": 13312
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.62666851415639,
|
| 599 |
+
"grad_norm": 0.09665919840335846,
|
| 600 |
+
"learning_rate": 0.0003194790303463687,
|
| 601 |
+
"loss": 0.016629021614789963,
|
| 602 |
+
"step": 13568
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.6384924483857558,
|
| 606 |
+
"grad_norm": 0.16131311655044556,
|
| 607 |
+
"learning_rate": 0.00030187420020572406,
|
| 608 |
+
"loss": 0.016964538022875786,
|
| 609 |
+
"step": 13824
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.6503163826151217,
|
| 613 |
+
"grad_norm": 0.18972595036029816,
|
| 614 |
+
"learning_rate": 0.00028455611743603626,
|
| 615 |
+
"loss": 0.01751314289867878,
|
| 616 |
+
"step": 14080
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.6621403168444876,
|
| 620 |
+
"grad_norm": 0.054903462529182434,
|
| 621 |
+
"learning_rate": 0.0002675498464898373,
|
| 622 |
+
"loss": 0.016607709228992462,
|
| 623 |
+
"step": 14336
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 0.6621403168444876,
|
| 627 |
+
"eval_cos_loss": 0.00858595758143889,
|
| 628 |
+
"eval_loss": 0.017996014871944983,
|
| 629 |
+
"eval_mse_loss": 0.017137419099634572,
|
| 630 |
+
"step": 14336
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.6621403168444876,
|
| 634 |
+
"eval_cos_loss": 0.00858595758143889,
|
| 635 |
+
"eval_loss": 0.017996014871944983,
|
| 636 |
+
"eval_mse_loss": 0.017137419099634572,
|
| 637 |
+
"eval_runtime": 48.1346,
|
| 638 |
+
"eval_samples_per_second": 581.556,
|
| 639 |
+
"eval_steps_per_second": 9.099,
|
| 640 |
+
"step": 14336
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.6739642510738534,
|
| 644 |
+
"grad_norm": 0.13527068495750427,
|
| 645 |
+
"learning_rate": 0.0002508800005345623,
|
| 646 |
+
"loss": 0.01666027121245861,
|
| 647 |
+
"step": 14592
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.6857881853032193,
|
| 651 |
+
"grad_norm": 0.11799178272485733,
|
| 652 |
+
"learning_rate": 0.00023457070582992562,
|
| 653 |
+
"loss": 0.01664375141263008,
|
| 654 |
+
"step": 14848
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.6976121195325851,
|
| 658 |
+
"grad_norm": 0.09718749672174454,
|
| 659 |
+
"learning_rate": 0.00021864556680999692,
|
| 660 |
+
"loss": 0.016424618661403656,
|
| 661 |
+
"step": 15104
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.709436053761951,
|
| 665 |
+
"grad_norm": 0.08593294024467468,
|
| 666 |
+
"learning_rate": 0.0002031276319205152,
|
| 667 |
+
"loss": 0.016002284362912178,
|
| 668 |
+
"step": 15360
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.709436053761951,
|
| 672 |
+
"eval_cos_loss": 0.007626088229263176,
|
| 673 |
+
"eval_loss": 0.01650830843915406,
|
| 674 |
+
"eval_mse_loss": 0.0157456995768379,
|
| 675 |
+
"step": 15360
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.709436053761951,
|
| 679 |
+
"eval_cos_loss": 0.007626088229263176,
|
| 680 |
+
"eval_loss": 0.01650830843915406,
|
| 681 |
+
"eval_mse_loss": 0.0157456995768379,
|
| 682 |
+
"eval_runtime": 48.0815,
|
| 683 |
+
"eval_samples_per_second": 582.198,
|
| 684 |
+
"eval_steps_per_second": 9.11,
|
| 685 |
+
"step": 15360
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.7212599879913169,
|
| 689 |
+
"grad_norm": 0.11202144622802734,
|
| 690 |
+
"learning_rate": 0.00018803936026088542,
|
| 691 |
+
"loss": 0.017690593376755714,
|
| 692 |
+
"step": 15616
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.7330839222206826,
|
| 696 |
+
"grad_norm": 0.08233276754617691,
|
| 697 |
+
"learning_rate": 0.00017340258907913464,
|
| 698 |
+
"loss": 0.016846561804413795,
|
| 699 |
+
"step": 15872
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.7449078564500485,
|
| 703 |
+
"grad_norm": 0.14655140042304993,
|
| 704 |
+
"learning_rate": 0.0001592385021668743,
|
| 705 |
+
"loss": 0.016557510942220688,
|
| 706 |
+
"step": 16128
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.7567317906794143,
|
| 710 |
+
"grad_norm": 0.06357849389314651,
|
| 711 |
+
"learning_rate": 0.0001455675992000087,
|
| 712 |
+
"loss": 0.016823401674628258,
|
| 713 |
+
"step": 16384
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.7567317906794143,
|
| 717 |
+
"eval_cos_loss": 0.0075487891793119205,
|
| 718 |
+
"eval_loss": 0.016473648955909425,
|
| 719 |
+
"eval_mse_loss": 0.01571877002312066,
|
| 720 |
+
"step": 16384
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.7567317906794143,
|
| 724 |
+
"eval_cos_loss": 0.0075487891793119205,
|
| 725 |
+
"eval_loss": 0.016473648955909425,
|
| 726 |
+
"eval_mse_loss": 0.01571877002312066,
|
| 727 |
+
"eval_runtime": 48.4424,
|
| 728 |
+
"eval_samples_per_second": 577.861,
|
| 729 |
+
"eval_steps_per_second": 9.042,
|
| 730 |
+
"step": 16384
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.7685557249087802,
|
| 734 |
+
"grad_norm": 0.07761925458908081,
|
| 735 |
+
"learning_rate": 0.000132409666069565,
|
| 736 |
+
"loss": 0.015980515629053116,
|
| 737 |
+
"step": 16640
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.780379659138146,
|
| 741 |
+
"grad_norm": 0.08687383681535721,
|
| 742 |
+
"learning_rate": 0.0001197837462455823,
|
| 743 |
+
"loss": 0.016285939142107964,
|
| 744 |
+
"step": 16896
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.7922035933675119,
|
| 748 |
+
"grad_norm": 0.05492519587278366,
|
| 749 |
+
"learning_rate": 0.00010770811321550749,
|
| 750 |
+
"loss": 0.017599251121282578,
|
| 751 |
+
"step": 17152
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.8040275275968778,
|
| 755 |
+
"grad_norm": 0.06340524554252625,
|
| 756 |
+
"learning_rate": 9.620024403698591e-05,
|
| 757 |
+
"loss": 0.016355257481336594,
|
| 758 |
+
"step": 17408
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.8040275275968778,
|
| 762 |
+
"eval_cos_loss": 0.007373531854001286,
|
| 763 |
+
"eval_loss": 0.015994041897654195,
|
| 764 |
+
"eval_mse_loss": 0.01525668867387422,
|
| 765 |
+
"step": 17408
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.8040275275968778,
|
| 769 |
+
"eval_cos_loss": 0.007373531854001286,
|
| 770 |
+
"eval_loss": 0.015994041897654195,
|
| 771 |
+
"eval_mse_loss": 0.01525668867387422,
|
| 772 |
+
"eval_runtime": 48.2951,
|
| 773 |
+
"eval_samples_per_second": 579.624,
|
| 774 |
+
"eval_steps_per_second": 9.069,
|
| 775 |
+
"step": 17408
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.8158514618262436,
|
| 779 |
+
"grad_norm": 0.05335954204201698,
|
| 780 |
+
"learning_rate": 8.527679404332429e-05,
|
| 781 |
+
"loss": 0.015534437261521816,
|
| 782 |
+
"step": 17664
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.8276753960556095,
|
| 786 |
+
"grad_norm": 0.04659634083509445,
|
| 787 |
+
"learning_rate": 7.495357273823544e-05,
|
| 788 |
+
"loss": 0.016477219760417938,
|
| 789 |
+
"step": 17920
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.8394993302849753,
|
| 793 |
+
"grad_norm": 0.06442078948020935,
|
| 794 |
+
"learning_rate": 6.524552091475183e-05,
|
| 795 |
+
"loss": 0.016562407836318016,
|
| 796 |
+
"step": 18176
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.8513232645143411,
|
| 800 |
+
"grad_norm": 0.09905882924795151,
|
| 801 |
+
"learning_rate": 5.6166689031422024e-05,
|
| 802 |
+
"loss": 0.0154965128749609,
|
| 803 |
+
"step": 18432
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.8513232645143411,
|
| 807 |
+
"eval_cos_loss": 0.007233276418899292,
|
| 808 |
+
"eval_loss": 0.015860445888874603,
|
| 809 |
+
"eval_mse_loss": 0.01513711828738451,
|
| 810 |
+
"step": 18432
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.8513232645143411,
|
| 814 |
+
"eval_cos_loss": 0.007233276418899292,
|
| 815 |
+
"eval_loss": 0.015860445888874603,
|
| 816 |
+
"eval_mse_loss": 0.01513711828738451,
|
| 817 |
+
"eval_runtime": 48.2642,
|
| 818 |
+
"eval_samples_per_second": 579.995,
|
| 819 |
+
"eval_steps_per_second": 9.075,
|
| 820 |
+
"step": 18432
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.8631471987437069,
|
| 824 |
+
"grad_norm": 0.0645645260810852,
|
| 825 |
+
"learning_rate": 4.773021687709067e-05,
|
| 826 |
+
"loss": 0.01556779071688652,
|
| 827 |
+
"step": 18688
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 0.8749711329730728,
|
| 831 |
+
"grad_norm": 0.04627804830670357,
|
| 832 |
+
"learning_rate": 3.994831455368719e-05,
|
| 833 |
+
"loss": 0.016697347164154053,
|
| 834 |
+
"step": 18944
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.8867950672024387,
|
| 838 |
+
"grad_norm": 0.07359793037176132,
|
| 839 |
+
"learning_rate": 3.283224480455282e-05,
|
| 840 |
+
"loss": 0.016459425911307335,
|
| 841 |
+
"step": 19200
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.8986190014318045,
|
| 845 |
+
"grad_norm": 0.07019244879484177,
|
| 846 |
+
"learning_rate": 2.639230671387627e-05,
|
| 847 |
+
"loss": 0.016459926962852478,
|
| 848 |
+
"step": 19456
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.8986190014318045,
|
| 852 |
+
"eval_cos_loss": 0.0070534555840021055,
|
| 853 |
+
"eval_loss": 0.015448579780510776,
|
| 854 |
+
"eval_mse_loss": 0.014743234202070118,
|
| 855 |
+
"step": 19456
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.8986190014318045,
|
| 859 |
+
"eval_cos_loss": 0.0070534555840021055,
|
| 860 |
+
"eval_loss": 0.015448579780510776,
|
| 861 |
+
"eval_mse_loss": 0.014743234202070118,
|
| 862 |
+
"eval_runtime": 49.0687,
|
| 863 |
+
"eval_samples_per_second": 570.486,
|
| 864 |
+
"eval_steps_per_second": 8.926,
|
| 865 |
+
"step": 19456
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.9104429356611704,
|
| 869 |
+
"grad_norm": 0.04372810944914818,
|
| 870 |
+
"learning_rate": 2.063782080083576e-05,
|
| 871 |
+
"loss": 0.015928251668810844,
|
| 872 |
+
"step": 19712
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.9222668698905362,
|
| 876 |
+
"grad_norm": 0.04366910457611084,
|
| 877 |
+
"learning_rate": 1.557711553001523e-05,
|
| 878 |
+
"loss": 0.01521321851760149,
|
| 879 |
+
"step": 19968
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.9340908041199021,
|
| 883 |
+
"grad_norm": 0.022554531693458557,
|
| 884 |
+
"learning_rate": 1.1217515257622269e-05,
|
| 885 |
+
"loss": 0.015970518812537193,
|
| 886 |
+
"step": 20224
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 0.945914738349268,
|
| 890 |
+
"grad_norm": 0.01744404435157776,
|
| 891 |
+
"learning_rate": 7.565329630950746e-06,
|
| 892 |
+
"loss": 0.016017938032746315,
|
| 893 |
+
"step": 20480
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 0.945914738349268,
|
| 897 |
+
"eval_cos_loss": 0.00700434575444215,
|
| 898 |
+
"eval_loss": 0.015360259538150678,
|
| 899 |
+
"eval_mse_loss": 0.01465982497674009,
|
| 900 |
+
"step": 20480
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 0.945914738349268,
|
| 904 |
+
"eval_cos_loss": 0.00700434575444215,
|
| 905 |
+
"eval_loss": 0.015360259538150678,
|
| 906 |
+
"eval_mse_loss": 0.01465982497674009,
|
| 907 |
+
"eval_runtime": 48.4748,
|
| 908 |
+
"eval_samples_per_second": 577.475,
|
| 909 |
+
"eval_steps_per_second": 9.036,
|
| 910 |
+
"step": 20480
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.9577386725786338,
|
| 914 |
+
"grad_norm": 0.04171235114336014,
|
| 915 |
+
"learning_rate": 4.62584445643166e-06,
|
| 916 |
+
"loss": 0.015838587656617165,
|
| 917 |
+
"step": 20736
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.9695626068079997,
|
| 921 |
+
"grad_norm": 0.04501587897539139,
|
| 922 |
+
"learning_rate": 2.40331404948807e-06,
|
| 923 |
+
"loss": 0.016174525022506714,
|
| 924 |
+
"step": 20992
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.9813865410373654,
|
| 928 |
+
"grad_norm": 0.017538242042064667,
|
| 929 |
+
"learning_rate": 9.009550772663965e-07,
|
| 930 |
+
"loss": 0.015945816412568092,
|
| 931 |
+
"step": 21248
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 0.9932104752667313,
|
| 935 |
+
"grad_norm": 0.03136038780212402,
|
| 936 |
+
"learning_rate": 1.2094190315575791e-07,
|
| 937 |
+
"loss": 0.015986243262887,
|
| 938 |
+
"step": 21504
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.9932104752667313,
|
| 942 |
+
"eval_cos_loss": 0.008081014438736418,
|
| 943 |
+
"eval_loss": 0.017069151780257624,
|
| 944 |
+
"eval_mse_loss": 0.01626105037237173,
|
| 945 |
+
"step": 21504
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.9932104752667313,
|
| 949 |
+
"eval_cos_loss": 0.008081014438736418,
|
| 950 |
+
"eval_loss": 0.017069151780257624,
|
| 951 |
+
"eval_mse_loss": 0.01626105037237173,
|
| 952 |
+
"eval_runtime": 48.4869,
|
| 953 |
+
"eval_samples_per_second": 577.331,
|
| 954 |
+
"eval_steps_per_second": 9.033,
|
| 955 |
+
"step": 21504
|
| 956 |
+
}
|
| 957 |
+
],
|
| 958 |
+
"logging_steps": 256,
|
| 959 |
+
"max_steps": 21651,
|
| 960 |
+
"num_input_tokens_seen": 0,
|
| 961 |
+
"num_train_epochs": 1,
|
| 962 |
+
"save_steps": 1024,
|
| 963 |
+
"stateful_callbacks": {
|
| 964 |
+
"TrainerControl": {
|
| 965 |
+
"args": {
|
| 966 |
+
"should_epoch_stop": false,
|
| 967 |
+
"should_evaluate": false,
|
| 968 |
+
"should_log": false,
|
| 969 |
+
"should_save": true,
|
| 970 |
+
"should_training_stop": false
|
| 971 |
+
},
|
| 972 |
+
"attributes": {}
|
| 973 |
+
}
|
| 974 |
+
},
|
| 975 |
+
"total_flos": 0.0,
|
| 976 |
+
"train_batch_size": 64,
|
| 977 |
+
"trial_name": null,
|
| 978 |
+
"trial_params": null
|
| 979 |
+
}
|
checkpoints-v6.0-discrete/checkpoint-21504/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61a16bb839f687414b8e48611327c4b9cfddeefe38c031ca70808f9a97c476b7
|
| 3 |
+
size 5137
|