add samelayer_reduce qwen3-4b-kto
Browse files- v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/activations/activations.pt +3 -0
- v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/checkpoints/final.pt +3 -0
- v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/metrics/training_metrics.json +80 -0
- v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/run_meta.json +34 -0
v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/activations/activations.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:50412b4c7873273edb4b1365b877094bd474b150a92db38962331fd4b88b16ca
|
| 3 |
+
size 1252425669
|
v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/checkpoints/final.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79e71633eb02aff4d557253684b9e5cad0b5f2571c768ca92e4ab2a230da2c7c
|
| 3 |
+
size 2517091162
|
v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/metrics/training_metrics.json
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epochs": [
|
| 3 |
+
1,
|
| 4 |
+
2,
|
| 5 |
+
3,
|
| 6 |
+
4
|
| 7 |
+
],
|
| 8 |
+
"train_loss": [
|
| 9 |
+
5.128696043520592,
|
| 10 |
+
2.698074368239145,
|
| 11 |
+
1.9589535239768598,
|
| 12 |
+
1.6976813955468457
|
| 13 |
+
],
|
| 14 |
+
"val_loss": [
|
| 15 |
+
3.2481538450530687,
|
| 16 |
+
2.153272160684875,
|
| 17 |
+
1.6496683384111415,
|
| 18 |
+
1.5583075802987782
|
| 19 |
+
],
|
| 20 |
+
"train_fve_base": [
|
| 21 |
+
0.7659889680561237,
|
| 22 |
+
0.8822093652621111,
|
| 23 |
+
0.9012698801859595,
|
| 24 |
+
0.9116996667987759
|
| 25 |
+
],
|
| 26 |
+
"train_fve_aligned": [
|
| 27 |
+
0.7795407430372667,
|
| 28 |
+
0.891486782604536,
|
| 29 |
+
0.9098470497047838,
|
| 30 |
+
0.9197595569033826
|
| 31 |
+
],
|
| 32 |
+
"val_fve_base": [
|
| 33 |
+
0.8671354732588323,
|
| 34 |
+
0.8875821838204149,
|
| 35 |
+
0.8955767476121793,
|
| 36 |
+
0.8975027883864198
|
| 37 |
+
],
|
| 38 |
+
"val_fve_aligned": [
|
| 39 |
+
0.8772461467388413,
|
| 40 |
+
0.8973232726152031,
|
| 41 |
+
0.9048313489135024,
|
| 42 |
+
0.9066294896665044
|
| 43 |
+
],
|
| 44 |
+
"dead_neurons": [
|
| 45 |
+
0.9232909989788458,
|
| 46 |
+
0.9566701101546947,
|
| 47 |
+
0.9575561995194519,
|
| 48 |
+
0.9573373428855245
|
| 49 |
+
],
|
| 50 |
+
"l0_base": [
|
| 51 |
+
229.3560091943958,
|
| 52 |
+
216.9810639229422,
|
| 53 |
+
215.1935566258027,
|
| 54 |
+
214.44505253940454
|
| 55 |
+
],
|
| 56 |
+
"l0_aligned": [
|
| 57 |
+
227.8894483362522,
|
| 58 |
+
216.73254159369526,
|
| 59 |
+
214.9229969352014,
|
| 60 |
+
214.1355078809107
|
| 61 |
+
],
|
| 62 |
+
"self_recon": [
|
| 63 |
+
3.336422686401474,
|
| 64 |
+
1.6159900503762619,
|
| 65 |
+
1.0519533833076713,
|
| 66 |
+
0.8389521715450342
|
| 67 |
+
],
|
| 68 |
+
"cross_recon": [
|
| 69 |
+
3.784384382585302,
|
| 70 |
+
2.3081956570495032,
|
| 71 |
+
1.8486120180643864,
|
| 72 |
+
1.7151102538840923
|
| 73 |
+
],
|
| 74 |
+
"sparsity": [
|
| 75 |
+
0.2785196033438534,
|
| 76 |
+
0.15880604364495493,
|
| 77 |
+
0.16755532944369442,
|
| 78 |
+
0.17268510582115798
|
| 79 |
+
]
|
| 80 |
+
}
|
v3-samelayer-seed-99/samelayer_reduce/qwen3-4b-kto/L23-25/run_meta.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model": "Qwen/Qwen3-4B-Instruct-2507",
|
| 3 |
+
"aligned_run_id": "qwen3-4b-kto",
|
| 4 |
+
"position": "last_prompt",
|
| 5 |
+
"multi_type": "reduce",
|
| 6 |
+
"source_activations": "/home/igarg/research/interp-alignment/interp_utils/crosscoder/results-multi-v1/v1/assembled_activations/qwen3-4b-kto/activations/activations.pt",
|
| 7 |
+
"base_activations_path": "/home/igarg/research/interp-alignment/interp_utils/crosscoder/results-multi-v1/v1/assembled_activations/qwen3-4b-kto/activations/activations.pt",
|
| 8 |
+
"source_layers": [
|
| 9 |
+
23,
|
| 10 |
+
24,
|
| 11 |
+
25
|
| 12 |
+
],
|
| 13 |
+
"base_source_layers": [
|
| 14 |
+
23,
|
| 15 |
+
24,
|
| 16 |
+
25
|
| 17 |
+
],
|
| 18 |
+
"layers": [
|
| 19 |
+
23,
|
| 20 |
+
24,
|
| 21 |
+
25
|
| 22 |
+
],
|
| 23 |
+
"base_layers": [
|
| 24 |
+
23,
|
| 25 |
+
24,
|
| 26 |
+
25
|
| 27 |
+
],
|
| 28 |
+
"aligned_layers": [
|
| 29 |
+
23,
|
| 30 |
+
24,
|
| 31 |
+
25
|
| 32 |
+
],
|
| 33 |
+
"layer_reduction": "mean"
|
| 34 |
+
}
|