Add Blockwise-OAT baseline artifacts (benchmarks/verify_training.json)
Browse files
benchmarks/verify_training.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checks": {
|
| 3 |
+
"overfit_ok": true,
|
| 4 |
+
"reference_loss_ok": true,
|
| 5 |
+
"refine_numerics_ok": true
|
| 6 |
+
},
|
| 7 |
+
"config": {
|
| 8 |
+
"batch_size": 32,
|
| 9 |
+
"d_model": 128,
|
| 10 |
+
"device": "cuda:0",
|
| 11 |
+
"n_tail": 4,
|
| 12 |
+
"overfit_steps": 120,
|
| 13 |
+
"prefix_len": 4,
|
| 14 |
+
"refine_iters": 2,
|
| 15 |
+
"vocab_size": 64
|
| 16 |
+
},
|
| 17 |
+
"overfit": {
|
| 18 |
+
"final_loss": 0.0031353291124105453,
|
| 19 |
+
"initial_loss": 4.357937812805176,
|
| 20 |
+
"loss_ratio": 0.0007194524674486704,
|
| 21 |
+
"tail_token_accuracy": 1.0
|
| 22 |
+
},
|
| 23 |
+
"passed": true,
|
| 24 |
+
"real_batch_sanity": null,
|
| 25 |
+
"reference_loss": {
|
| 26 |
+
"abs_diff": 0.0,
|
| 27 |
+
"manual_loss": 4.265489101409912,
|
| 28 |
+
"training_loss": 4.265489101409912
|
| 29 |
+
},
|
| 30 |
+
"refine_stability": {
|
| 31 |
+
"changed_token_fraction_iter2_vs_iter1": 0.1875,
|
| 32 |
+
"finite_logits": true,
|
| 33 |
+
"max_logit_delta": 0.19519612193107605,
|
| 34 |
+
"token_bounds_ok": true
|
| 35 |
+
}
|
| 36 |
+
}
|