Reference provenance files in model card
Browse files
README.md
CHANGED
|
@@ -83,3 +83,10 @@ Trained with [speculators](https://github.com/neuralmagic/speculators) for 3
|
|
| 83 |
epochs (`checkpoint_best` = best validation epoch). Validation at the selected
|
| 84 |
checkpoint: accept_len 3.67, accept_rate 0.40, pos-0 acc 0.776. Loss
|
| 85 |
`{"ce": 0.1, "tv": 0.9}`, lr 1e-4, seq len 8192, block size 16.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
epochs (`checkpoint_best` = best validation epoch). Validation at the selected
|
| 84 |
checkpoint: accept_len 3.67, accept_rate 0.40, pos-0 acc 0.776. Loss
|
| 85 |
`{"ce": 0.1, "tv": 0.9}`, lr 1e-4, seq len 8192, block size 16.
|
| 86 |
+
|
| 87 |
+
Provenance / reproducibility:
|
| 88 |
+
- [`train_command.txt`](./train_command.txt) — exact training command, git SHA, and
|
| 89 |
+
package versions (speculators 0.7.0.dev155, vllm 0.27.1, transformers 5.15.0,
|
| 90 |
+
torch 2.13.0+cu130, compressed-tensors 0.17.0).
|
| 91 |
+
- [`training_state.json`](./training_state.json) — epoch / global step of this checkpoint.
|
| 92 |
+
- [`val_metrics.json`](./val_metrics.json) — full validation metrics.
|