checkpoint-7000
Browse files- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/README.md +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/adapter_config.json +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/adapter_model.safetensors +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/optimizer.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/rng_state.pth +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/scheduler.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/trainer_state.json +2208 -3
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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 +9 -9
- 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-4000 → checkpoint-7000}/README.md
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/adapter_config.json
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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:f54ae018eb42047205f34a408438b5b613c2797fd402314a2e83a16e2d0c246b
|
| 3 |
size 134604152
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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 sha256:84cf35a2924f44a22c4ad3b7e7c91dfbc3a36aeadd20193df87e7cda7e1a0b6e
|
| 3 |
size 269426431
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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 sha256:838f63b2cc3d5c3dd24c183a6a575427c7e9f25a9457b4a6a917bea681e36f49
|
| 3 |
size 14645
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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 sha256:7774859a19e9f0ab6a5319bc0fe9826f553038c2dc3f106399331bc680d8323a
|
| 3 |
size 1465
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/trainer_state.json
RENAMED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 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,
|
|
@@ -2913,6 +2913,2211 @@
|
|
| 2913 |
"learning_rate": 6.750094446543257e-05,
|
| 2914 |
"loss": 0.9594356536865234,
|
| 2915 |
"step": 4000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2916 |
}
|
| 2917 |
],
|
| 2918 |
"logging_steps": 10,
|
|
@@ -2932,7 +5137,7 @@
|
|
| 2932 |
"attributes": {}
|
| 2933 |
}
|
| 2934 |
},
|
| 2935 |
-
"total_flos":
|
| 2936 |
"train_batch_size": 2,
|
| 2937 |
"trial_name": null,
|
| 2938 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.256096193104067,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 7000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2913 |
"learning_rate": 6.750094446543257e-05,
|
| 2914 |
"loss": 0.9594356536865234,
|
| 2915 |
"step": 4000
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.717858985575521,
|
| 2919 |
+
"step": 4000,
|
| 2920 |
+
"test_entity_exact_match": 0.0,
|
| 2921 |
+
"test_entity_fix_rate": 0.23646554153489682,
|
| 2922 |
+
"test_entity_fix_rate_lenient": 0.22922824302134648,
|
| 2923 |
+
"test_entity_fix_rate_recoverable": 0.23269956895111438,
|
| 2924 |
+
"test_entity_hallucination_rate": 0.13267326732673268,
|
| 2925 |
+
"test_entity_preservation_rate": 0.6708058549139456,
|
| 2926 |
+
"test_entity_preservation_rate_lenient": 0.6731650051961049,
|
| 2927 |
+
"test_entity_targeted_score": 0.45363569822442124,
|
| 2928 |
+
"test_entity_targeted_score_lenient": 0.4511966241087257,
|
| 2929 |
+
"test_entity_wer": 0.5756076472283601,
|
| 2930 |
+
"test_entity_wer_zwnj_normalized": 0.5715032930903812
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.717858985575521,
|
| 2934 |
+
"step": 4000,
|
| 2935 |
+
"test_typo_exact_match": 0.0,
|
| 2936 |
+
"test_typo_fix_rate": 0.31737107405971304,
|
| 2937 |
+
"test_typo_fix_rate_lenient": 0.31215069343821156,
|
| 2938 |
+
"test_typo_fix_rate_recoverable": 0.316099201345103,
|
| 2939 |
+
"test_typo_hallucination_rate": 0.17039106145251395,
|
| 2940 |
+
"test_typo_preservation_rate": 0.786258195395878,
|
| 2941 |
+
"test_typo_preservation_rate_lenient": 0.7890943186146678,
|
| 2942 |
+
"test_typo_targeted_score": 0.5518146347277956,
|
| 2943 |
+
"test_typo_targeted_score_lenient": 0.5506225060264397,
|
| 2944 |
+
"test_typo_wer": 0.5879860100584616,
|
| 2945 |
+
"test_typo_wer_zwnj_normalized": 0.5820122029052105
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 0.717858985575521,
|
| 2949 |
+
"step": 4000,
|
| 2950 |
+
"test_best_wer": 0.2231321104545041
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.7196536330394598,
|
| 2954 |
+
"grad_norm": 0.8268827795982361,
|
| 2955 |
+
"learning_rate": 6.740649792217605e-05,
|
| 2956 |
+
"loss": 0.7053637981414795,
|
| 2957 |
+
"step": 4010
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.7214482805033986,
|
| 2961 |
+
"grad_norm": 1.0276139974594116,
|
| 2962 |
+
"learning_rate": 6.731205137891953e-05,
|
| 2963 |
+
"loss": 0.7204613208770752,
|
| 2964 |
+
"step": 4020
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.7232429279673375,
|
| 2968 |
+
"grad_norm": 1.1466572284698486,
|
| 2969 |
+
"learning_rate": 6.721760483566302e-05,
|
| 2970 |
+
"loss": 0.6874489784240723,
|
| 2971 |
+
"step": 4030
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.7250375754312762,
|
| 2975 |
+
"grad_norm": 1.6566600799560547,
|
| 2976 |
+
"learning_rate": 6.71231582924065e-05,
|
| 2977 |
+
"loss": 0.8332732200622559,
|
| 2978 |
+
"step": 4040
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.726832222895215,
|
| 2982 |
+
"grad_norm": 3.702282190322876,
|
| 2983 |
+
"learning_rate": 6.702871174914998e-05,
|
| 2984 |
+
"loss": 0.8752366065979004,
|
| 2985 |
+
"step": 4050
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.7286268703591539,
|
| 2989 |
+
"grad_norm": 0.7897398471832275,
|
| 2990 |
+
"learning_rate": 6.693426520589347e-05,
|
| 2991 |
+
"loss": 0.7186062335968018,
|
| 2992 |
+
"step": 4060
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.7304215178230926,
|
| 2996 |
+
"grad_norm": 0.9881529211997986,
|
| 2997 |
+
"learning_rate": 6.683981866263695e-05,
|
| 2998 |
+
"loss": 0.731151008605957,
|
| 2999 |
+
"step": 4070
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.7322161652870315,
|
| 3003 |
+
"grad_norm": 1.118485927581787,
|
| 3004 |
+
"learning_rate": 6.674537211938043e-05,
|
| 3005 |
+
"loss": 0.7113176822662354,
|
| 3006 |
+
"step": 4080
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.7340108127509702,
|
| 3010 |
+
"grad_norm": 1.6644984483718872,
|
| 3011 |
+
"learning_rate": 6.665092557612392e-05,
|
| 3012 |
+
"loss": 0.8222126007080078,
|
| 3013 |
+
"step": 4090
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 0.735805460214909,
|
| 3017 |
+
"grad_norm": 2.4951610565185547,
|
| 3018 |
+
"learning_rate": 6.65564790328674e-05,
|
| 3019 |
+
"loss": 0.9295941352844238,
|
| 3020 |
+
"step": 4100
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.7376001076788479,
|
| 3024 |
+
"grad_norm": 0.8182881474494934,
|
| 3025 |
+
"learning_rate": 6.646203248961088e-05,
|
| 3026 |
+
"loss": 0.7256208419799804,
|
| 3027 |
+
"step": 4110
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 0.7393947551427866,
|
| 3031 |
+
"grad_norm": 0.9188361763954163,
|
| 3032 |
+
"learning_rate": 6.636758594635437e-05,
|
| 3033 |
+
"loss": 0.7089800357818603,
|
| 3034 |
+
"step": 4120
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.7411894026067254,
|
| 3038 |
+
"grad_norm": 1.1344740390777588,
|
| 3039 |
+
"learning_rate": 6.627313940309785e-05,
|
| 3040 |
+
"loss": 0.6811440467834473,
|
| 3041 |
+
"step": 4130
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.7429840500706643,
|
| 3045 |
+
"grad_norm": 1.3251533508300781,
|
| 3046 |
+
"learning_rate": 6.617869285984133e-05,
|
| 3047 |
+
"loss": 0.8110069274902344,
|
| 3048 |
+
"step": 4140
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.744778697534603,
|
| 3052 |
+
"grad_norm": 4.748185157775879,
|
| 3053 |
+
"learning_rate": 6.608424631658482e-05,
|
| 3054 |
+
"loss": 1.0361065864562988,
|
| 3055 |
+
"step": 4150
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.7465733449985419,
|
| 3059 |
+
"grad_norm": 0.8669788837432861,
|
| 3060 |
+
"learning_rate": 6.59897997733283e-05,
|
| 3061 |
+
"loss": 0.7604600429534912,
|
| 3062 |
+
"step": 4160
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.7483679924624806,
|
| 3066 |
+
"grad_norm": 0.8855018019676208,
|
| 3067 |
+
"learning_rate": 6.589535323007178e-05,
|
| 3068 |
+
"loss": 0.712589168548584,
|
| 3069 |
+
"step": 4170
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 0.7501626399264194,
|
| 3073 |
+
"grad_norm": 1.2251121997833252,
|
| 3074 |
+
"learning_rate": 6.580090668681527e-05,
|
| 3075 |
+
"loss": 0.7184146404266357,
|
| 3076 |
+
"step": 4180
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 0.7519572873903583,
|
| 3080 |
+
"grad_norm": 1.4125711917877197,
|
| 3081 |
+
"learning_rate": 6.570646014355875e-05,
|
| 3082 |
+
"loss": 0.8462773323059082,
|
| 3083 |
+
"step": 4190
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 0.753751934854297,
|
| 3087 |
+
"grad_norm": 3.009613275527954,
|
| 3088 |
+
"learning_rate": 6.561201360030223e-05,
|
| 3089 |
+
"loss": 0.9116678237915039,
|
| 3090 |
+
"step": 4200
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 0.7555465823182359,
|
| 3094 |
+
"grad_norm": 0.7657504081726074,
|
| 3095 |
+
"learning_rate": 6.551756705704572e-05,
|
| 3096 |
+
"loss": 0.7391473770141601,
|
| 3097 |
+
"step": 4210
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 0.7573412297821747,
|
| 3101 |
+
"grad_norm": 0.9948172569274902,
|
| 3102 |
+
"learning_rate": 6.542312051378919e-05,
|
| 3103 |
+
"loss": 0.7430819034576416,
|
| 3104 |
+
"step": 4220
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.7591358772461134,
|
| 3108 |
+
"grad_norm": 1.3323181867599487,
|
| 3109 |
+
"learning_rate": 6.532867397053269e-05,
|
| 3110 |
+
"loss": 0.7868825435638428,
|
| 3111 |
+
"step": 4230
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 0.7609305247100523,
|
| 3115 |
+
"grad_norm": 1.4949253797531128,
|
| 3116 |
+
"learning_rate": 6.523422742727617e-05,
|
| 3117 |
+
"loss": 0.8261573791503907,
|
| 3118 |
+
"step": 4240
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 0.762725172173991,
|
| 3122 |
+
"grad_norm": 4.123211860656738,
|
| 3123 |
+
"learning_rate": 6.513978088401965e-05,
|
| 3124 |
+
"loss": 0.9821381568908691,
|
| 3125 |
+
"step": 4250
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 0.7645198196379299,
|
| 3129 |
+
"grad_norm": 0.8264675140380859,
|
| 3130 |
+
"learning_rate": 6.504533434076314e-05,
|
| 3131 |
+
"loss": 0.7165445327758789,
|
| 3132 |
+
"step": 4260
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 0.7663144671018687,
|
| 3136 |
+
"grad_norm": 0.9300968050956726,
|
| 3137 |
+
"learning_rate": 6.49508877975066e-05,
|
| 3138 |
+
"loss": 0.7628948211669921,
|
| 3139 |
+
"step": 4270
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 0.7681091145658074,
|
| 3143 |
+
"grad_norm": 1.1202552318572998,
|
| 3144 |
+
"learning_rate": 6.48564412542501e-05,
|
| 3145 |
+
"loss": 0.7190794944763184,
|
| 3146 |
+
"step": 4280
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.7699037620297463,
|
| 3150 |
+
"grad_norm": 1.514113187789917,
|
| 3151 |
+
"learning_rate": 6.476199471099357e-05,
|
| 3152 |
+
"loss": 0.8148799896240234,
|
| 3153 |
+
"step": 4290
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 0.7716984094936851,
|
| 3157 |
+
"grad_norm": 3.2658019065856934,
|
| 3158 |
+
"learning_rate": 6.466754816773707e-05,
|
| 3159 |
+
"loss": 0.938233757019043,
|
| 3160 |
+
"step": 4300
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 0.7734930569576239,
|
| 3164 |
+
"grad_norm": 0.9332228302955627,
|
| 3165 |
+
"learning_rate": 6.457310162448055e-05,
|
| 3166 |
+
"loss": 0.7350199699401856,
|
| 3167 |
+
"step": 4310
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 0.7752877044215627,
|
| 3171 |
+
"grad_norm": 0.9837442636489868,
|
| 3172 |
+
"learning_rate": 6.447865508122402e-05,
|
| 3173 |
+
"loss": 0.7245872974395752,
|
| 3174 |
+
"step": 4320
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 0.7770823518855015,
|
| 3178 |
+
"grad_norm": 1.4489710330963135,
|
| 3179 |
+
"learning_rate": 6.438420853796752e-05,
|
| 3180 |
+
"loss": 0.7148647308349609,
|
| 3181 |
+
"step": 4330
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 0.7788769993494403,
|
| 3185 |
+
"grad_norm": 1.4325411319732666,
|
| 3186 |
+
"learning_rate": 6.428976199471099e-05,
|
| 3187 |
+
"loss": 0.7869903087615967,
|
| 3188 |
+
"step": 4340
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 0.7806716468133791,
|
| 3192 |
+
"grad_norm": 4.452845096588135,
|
| 3193 |
+
"learning_rate": 6.419531545145449e-05,
|
| 3194 |
+
"loss": 0.8885753631591797,
|
| 3195 |
+
"step": 4350
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 0.782466294277318,
|
| 3199 |
+
"grad_norm": 0.8960409760475159,
|
| 3200 |
+
"learning_rate": 6.410086890819797e-05,
|
| 3201 |
+
"loss": 0.692021369934082,
|
| 3202 |
+
"step": 4360
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 0.7842609417412567,
|
| 3206 |
+
"grad_norm": 0.9574803113937378,
|
| 3207 |
+
"learning_rate": 6.400642236494144e-05,
|
| 3208 |
+
"loss": 0.7156808376312256,
|
| 3209 |
+
"step": 4370
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 0.7860555892051955,
|
| 3213 |
+
"grad_norm": 1.0848815441131592,
|
| 3214 |
+
"learning_rate": 6.391197582168494e-05,
|
| 3215 |
+
"loss": 0.6753977775573731,
|
| 3216 |
+
"step": 4380
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 0.7878502366691343,
|
| 3220 |
+
"grad_norm": 1.3776320219039917,
|
| 3221 |
+
"learning_rate": 6.38175292784284e-05,
|
| 3222 |
+
"loss": 0.9061891555786132,
|
| 3223 |
+
"step": 4390
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 0.7896448841330731,
|
| 3227 |
+
"grad_norm": 3.7691376209259033,
|
| 3228 |
+
"learning_rate": 6.372308273517189e-05,
|
| 3229 |
+
"loss": 0.9163275718688965,
|
| 3230 |
+
"step": 4400
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 0.7914395315970119,
|
| 3234 |
+
"grad_norm": 0.8445814847946167,
|
| 3235 |
+
"learning_rate": 6.362863619191539e-05,
|
| 3236 |
+
"loss": 0.73052978515625,
|
| 3237 |
+
"step": 4410
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 0.7932341790609507,
|
| 3241 |
+
"grad_norm": 0.9680049419403076,
|
| 3242 |
+
"learning_rate": 6.353418964865886e-05,
|
| 3243 |
+
"loss": 0.7528566837310791,
|
| 3244 |
+
"step": 4420
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 0.7950288265248895,
|
| 3248 |
+
"grad_norm": 1.2166266441345215,
|
| 3249 |
+
"learning_rate": 6.343974310540235e-05,
|
| 3250 |
+
"loss": 0.7165881156921386,
|
| 3251 |
+
"step": 4430
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 0.7968234739888284,
|
| 3255 |
+
"grad_norm": 1.4934667348861694,
|
| 3256 |
+
"learning_rate": 6.334529656214582e-05,
|
| 3257 |
+
"loss": 0.7936708927154541,
|
| 3258 |
+
"step": 4440
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 0.7986181214527671,
|
| 3262 |
+
"grad_norm": 3.400346040725708,
|
| 3263 |
+
"learning_rate": 6.32508500188893e-05,
|
| 3264 |
+
"loss": 0.9394660949707031,
|
| 3265 |
+
"step": 4450
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 0.8004127689167059,
|
| 3269 |
+
"grad_norm": 0.8123710751533508,
|
| 3270 |
+
"learning_rate": 6.31564034756328e-05,
|
| 3271 |
+
"loss": 0.6932941436767578,
|
| 3272 |
+
"step": 4460
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 0.8022074163806447,
|
| 3276 |
+
"grad_norm": 0.986383318901062,
|
| 3277 |
+
"learning_rate": 6.306195693237627e-05,
|
| 3278 |
+
"loss": 0.7004058361053467,
|
| 3279 |
+
"step": 4470
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 0.8040020638445835,
|
| 3283 |
+
"grad_norm": 1.1018342971801758,
|
| 3284 |
+
"learning_rate": 6.296751038911977e-05,
|
| 3285 |
+
"loss": 0.7086675643920899,
|
| 3286 |
+
"step": 4480
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 0.8057967113085224,
|
| 3290 |
+
"grad_norm": 1.3698561191558838,
|
| 3291 |
+
"learning_rate": 6.287306384586324e-05,
|
| 3292 |
+
"loss": 0.8011689186096191,
|
| 3293 |
+
"step": 4490
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 0.8075913587724611,
|
| 3297 |
+
"grad_norm": 3.463945150375366,
|
| 3298 |
+
"learning_rate": 6.277861730260672e-05,
|
| 3299 |
+
"loss": 0.9159812927246094,
|
| 3300 |
+
"step": 4500
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 0.8093860062363999,
|
| 3304 |
+
"grad_norm": 0.8161361217498779,
|
| 3305 |
+
"learning_rate": 6.268417075935021e-05,
|
| 3306 |
+
"loss": 0.6879207134246826,
|
| 3307 |
+
"step": 4510
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 0.8111806537003388,
|
| 3311 |
+
"grad_norm": 1.03386390209198,
|
| 3312 |
+
"learning_rate": 6.258972421609369e-05,
|
| 3313 |
+
"loss": 0.7262889385223389,
|
| 3314 |
+
"step": 4520
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 0.8129753011642775,
|
| 3318 |
+
"grad_norm": 1.2165919542312622,
|
| 3319 |
+
"learning_rate": 6.249527767283719e-05,
|
| 3320 |
+
"loss": 0.7320633411407471,
|
| 3321 |
+
"step": 4530
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 0.8147699486282164,
|
| 3325 |
+
"grad_norm": 1.5375968217849731,
|
| 3326 |
+
"learning_rate": 6.240083112958066e-05,
|
| 3327 |
+
"loss": 0.7480690956115723,
|
| 3328 |
+
"step": 4540
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 0.8165645960921551,
|
| 3332 |
+
"grad_norm": 5.311518669128418,
|
| 3333 |
+
"learning_rate": 6.230638458632414e-05,
|
| 3334 |
+
"loss": 0.947059154510498,
|
| 3335 |
+
"step": 4550
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 0.8183592435560939,
|
| 3339 |
+
"grad_norm": 0.874262273311615,
|
| 3340 |
+
"learning_rate": 6.221193804306762e-05,
|
| 3341 |
+
"loss": 0.7604807853698731,
|
| 3342 |
+
"step": 4560
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 0.8201538910200328,
|
| 3346 |
+
"grad_norm": 0.949961245059967,
|
| 3347 |
+
"learning_rate": 6.211749149981111e-05,
|
| 3348 |
+
"loss": 0.715032958984375,
|
| 3349 |
+
"step": 4570
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 0.8219485384839715,
|
| 3353 |
+
"grad_norm": 1.1289318799972534,
|
| 3354 |
+
"learning_rate": 6.202304495655459e-05,
|
| 3355 |
+
"loss": 0.6810876846313476,
|
| 3356 |
+
"step": 4580
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 0.8237431859479104,
|
| 3360 |
+
"grad_norm": 1.4804030656814575,
|
| 3361 |
+
"learning_rate": 6.192859841329807e-05,
|
| 3362 |
+
"loss": 0.7983368873596192,
|
| 3363 |
+
"step": 4590
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 0.8255378334118492,
|
| 3367 |
+
"grad_norm": 4.038904666900635,
|
| 3368 |
+
"learning_rate": 6.183415187004156e-05,
|
| 3369 |
+
"loss": 0.9614683151245117,
|
| 3370 |
+
"step": 4600
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 0.8273324808757879,
|
| 3374 |
+
"grad_norm": 0.8533964157104492,
|
| 3375 |
+
"learning_rate": 6.173970532678504e-05,
|
| 3376 |
+
"loss": 0.6545410633087159,
|
| 3377 |
+
"step": 4610
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 0.8291271283397268,
|
| 3381 |
+
"grad_norm": 0.9574486017227173,
|
| 3382 |
+
"learning_rate": 6.164525878352852e-05,
|
| 3383 |
+
"loss": 0.6657931327819824,
|
| 3384 |
+
"step": 4620
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 0.8309217758036656,
|
| 3388 |
+
"grad_norm": 1.1819963455200195,
|
| 3389 |
+
"learning_rate": 6.155081224027201e-05,
|
| 3390 |
+
"loss": 0.7103557109832763,
|
| 3391 |
+
"step": 4630
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 0.8327164232676044,
|
| 3395 |
+
"grad_norm": 1.3845491409301758,
|
| 3396 |
+
"learning_rate": 6.145636569701549e-05,
|
| 3397 |
+
"loss": 0.7800614356994628,
|
| 3398 |
+
"step": 4640
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 0.8345110707315432,
|
| 3402 |
+
"grad_norm": 3.9027099609375,
|
| 3403 |
+
"learning_rate": 6.136191915375898e-05,
|
| 3404 |
+
"loss": 0.9763243675231934,
|
| 3405 |
+
"step": 4650
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 0.8363057181954819,
|
| 3409 |
+
"grad_norm": 0.8903900384902954,
|
| 3410 |
+
"learning_rate": 6.126747261050246e-05,
|
| 3411 |
+
"loss": 0.6765525817871094,
|
| 3412 |
+
"step": 4660
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 0.8381003656594208,
|
| 3416 |
+
"grad_norm": 1.0684949159622192,
|
| 3417 |
+
"learning_rate": 6.117302606724594e-05,
|
| 3418 |
+
"loss": 0.7025310516357421,
|
| 3419 |
+
"step": 4670
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 0.8398950131233596,
|
| 3423 |
+
"grad_norm": 1.1708343029022217,
|
| 3424 |
+
"learning_rate": 6.107857952398943e-05,
|
| 3425 |
+
"loss": 0.7327500820159912,
|
| 3426 |
+
"step": 4680
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 0.8416896605872983,
|
| 3430 |
+
"grad_norm": 1.548041820526123,
|
| 3431 |
+
"learning_rate": 6.098413298073291e-05,
|
| 3432 |
+
"loss": 0.8242672920227051,
|
| 3433 |
+
"step": 4690
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 0.8434843080512372,
|
| 3437 |
+
"grad_norm": 4.831801891326904,
|
| 3438 |
+
"learning_rate": 6.088968643747639e-05,
|
| 3439 |
+
"loss": 0.9189159393310546,
|
| 3440 |
+
"step": 4700
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 0.845278955515176,
|
| 3444 |
+
"grad_norm": 0.7827987670898438,
|
| 3445 |
+
"learning_rate": 6.079523989421988e-05,
|
| 3446 |
+
"loss": 0.6926255702972413,
|
| 3447 |
+
"step": 4710
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 0.8470736029791148,
|
| 3451 |
+
"grad_norm": 1.0675493478775024,
|
| 3452 |
+
"learning_rate": 6.070079335096336e-05,
|
| 3453 |
+
"loss": 0.6892650604248047,
|
| 3454 |
+
"step": 4720
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 0.8488682504430536,
|
| 3458 |
+
"grad_norm": 1.4141958951950073,
|
| 3459 |
+
"learning_rate": 6.0606346807706836e-05,
|
| 3460 |
+
"loss": 0.7153830528259277,
|
| 3461 |
+
"step": 4730
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 0.8506628979069923,
|
| 3465 |
+
"grad_norm": 1.5664619207382202,
|
| 3466 |
+
"learning_rate": 6.0511900264450326e-05,
|
| 3467 |
+
"loss": 0.8375289916992188,
|
| 3468 |
+
"step": 4740
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 0.8524575453709312,
|
| 3472 |
+
"grad_norm": 4.490540027618408,
|
| 3473 |
+
"learning_rate": 6.041745372119381e-05,
|
| 3474 |
+
"loss": 0.8840279579162598,
|
| 3475 |
+
"step": 4750
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 0.85425219283487,
|
| 3479 |
+
"grad_norm": 0.8381492495536804,
|
| 3480 |
+
"learning_rate": 6.0323007177937286e-05,
|
| 3481 |
+
"loss": 0.6521424293518067,
|
| 3482 |
+
"step": 4760
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 0.8560468402988088,
|
| 3486 |
+
"grad_norm": 0.9900016784667969,
|
| 3487 |
+
"learning_rate": 6.0228560634680776e-05,
|
| 3488 |
+
"loss": 0.7005383491516113,
|
| 3489 |
+
"step": 4770
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 0.8578414877627476,
|
| 3493 |
+
"grad_norm": 1.2890015840530396,
|
| 3494 |
+
"learning_rate": 6.013411409142425e-05,
|
| 3495 |
+
"loss": 0.6942355632781982,
|
| 3496 |
+
"step": 4780
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 0.8596361352266864,
|
| 3500 |
+
"grad_norm": 1.5667506456375122,
|
| 3501 |
+
"learning_rate": 6.003966754816774e-05,
|
| 3502 |
+
"loss": 0.798072099685669,
|
| 3503 |
+
"step": 4790
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 0.8614307826906252,
|
| 3507 |
+
"grad_norm": 3.8494532108306885,
|
| 3508 |
+
"learning_rate": 5.994522100491122e-05,
|
| 3509 |
+
"loss": 0.959134578704834,
|
| 3510 |
+
"step": 4800
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 0.863225430154564,
|
| 3514 |
+
"grad_norm": 0.8042945265769958,
|
| 3515 |
+
"learning_rate": 5.98507744616547e-05,
|
| 3516 |
+
"loss": 0.7243553638458252,
|
| 3517 |
+
"step": 4810
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 0.8650200776185029,
|
| 3521 |
+
"grad_norm": 0.9480759501457214,
|
| 3522 |
+
"learning_rate": 5.975632791839819e-05,
|
| 3523 |
+
"loss": 0.6505447864532471,
|
| 3524 |
+
"step": 4820
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 0.8668147250824416,
|
| 3528 |
+
"grad_norm": 1.2419155836105347,
|
| 3529 |
+
"learning_rate": 5.966188137514167e-05,
|
| 3530 |
+
"loss": 0.7647704124450684,
|
| 3531 |
+
"step": 4830
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 0.8686093725463804,
|
| 3535 |
+
"grad_norm": 1.612558364868164,
|
| 3536 |
+
"learning_rate": 5.956743483188516e-05,
|
| 3537 |
+
"loss": 0.8548855781555176,
|
| 3538 |
+
"step": 4840
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 0.8704040200103192,
|
| 3542 |
+
"grad_norm": 4.58271598815918,
|
| 3543 |
+
"learning_rate": 5.947298828862864e-05,
|
| 3544 |
+
"loss": 0.843088436126709,
|
| 3545 |
+
"step": 4850
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 0.872198667474258,
|
| 3549 |
+
"grad_norm": 0.8638464212417603,
|
| 3550 |
+
"learning_rate": 5.937854174537212e-05,
|
| 3551 |
+
"loss": 0.649378776550293,
|
| 3552 |
+
"step": 4860
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 0.8739933149381969,
|
| 3556 |
+
"grad_norm": 0.8889257311820984,
|
| 3557 |
+
"learning_rate": 5.928409520211561e-05,
|
| 3558 |
+
"loss": 0.7091721057891845,
|
| 3559 |
+
"step": 4870
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 0.8757879624021356,
|
| 3563 |
+
"grad_norm": 1.238863468170166,
|
| 3564 |
+
"learning_rate": 5.918964865885909e-05,
|
| 3565 |
+
"loss": 0.7717015743255615,
|
| 3566 |
+
"step": 4880
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 0.8775826098660744,
|
| 3570 |
+
"grad_norm": 1.5579394102096558,
|
| 3571 |
+
"learning_rate": 5.909520211560258e-05,
|
| 3572 |
+
"loss": 0.7820851325988769,
|
| 3573 |
+
"step": 4890
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 0.8793772573300133,
|
| 3577 |
+
"grad_norm": 3.1934468746185303,
|
| 3578 |
+
"learning_rate": 5.9000755572346054e-05,
|
| 3579 |
+
"loss": 0.849330997467041,
|
| 3580 |
+
"step": 4900
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 0.881171904793952,
|
| 3584 |
+
"grad_norm": 0.8095167875289917,
|
| 3585 |
+
"learning_rate": 5.890630902908954e-05,
|
| 3586 |
+
"loss": 0.729648208618164,
|
| 3587 |
+
"step": 4910
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 0.8829665522578909,
|
| 3591 |
+
"grad_norm": 1.033992052078247,
|
| 3592 |
+
"learning_rate": 5.881186248583303e-05,
|
| 3593 |
+
"loss": 0.6961370944976807,
|
| 3594 |
+
"step": 4920
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 0.8847611997218296,
|
| 3598 |
+
"grad_norm": 1.2645893096923828,
|
| 3599 |
+
"learning_rate": 5.8717415942576504e-05,
|
| 3600 |
+
"loss": 0.7022522926330567,
|
| 3601 |
+
"step": 4930
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 0.8865558471857684,
|
| 3605 |
+
"grad_norm": 1.6214739084243774,
|
| 3606 |
+
"learning_rate": 5.862296939931998e-05,
|
| 3607 |
+
"loss": 0.7820470809936524,
|
| 3608 |
+
"step": 4940
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 0.8883504946497073,
|
| 3612 |
+
"grad_norm": 3.837054967880249,
|
| 3613 |
+
"learning_rate": 5.852852285606347e-05,
|
| 3614 |
+
"loss": 0.8870712280273437,
|
| 3615 |
+
"step": 4950
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 0.890145142113646,
|
| 3619 |
+
"grad_norm": 0.9028179049491882,
|
| 3620 |
+
"learning_rate": 5.8434076312806954e-05,
|
| 3621 |
+
"loss": 0.7104959964752198,
|
| 3622 |
+
"step": 4960
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 0.8919397895775849,
|
| 3626 |
+
"grad_norm": 0.9380175471305847,
|
| 3627 |
+
"learning_rate": 5.833962976955044e-05,
|
| 3628 |
+
"loss": 0.7519885540008545,
|
| 3629 |
+
"step": 4970
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 0.8937344370415237,
|
| 3633 |
+
"grad_norm": 1.1569966077804565,
|
| 3634 |
+
"learning_rate": 5.824518322629392e-05,
|
| 3635 |
+
"loss": 0.7281805992126464,
|
| 3636 |
+
"step": 4980
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 0.8955290845054624,
|
| 3640 |
+
"grad_norm": 1.584986686706543,
|
| 3641 |
+
"learning_rate": 5.81507366830374e-05,
|
| 3642 |
+
"loss": 0.8723097801208496,
|
| 3643 |
+
"step": 4990
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 0.8973237319694013,
|
| 3647 |
+
"grad_norm": 4.8088274002075195,
|
| 3648 |
+
"learning_rate": 5.805629013978089e-05,
|
| 3649 |
+
"loss": 0.9866933822631836,
|
| 3650 |
+
"step": 5000
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 0.8973237319694013,
|
| 3654 |
+
"step": 5000,
|
| 3655 |
+
"test_entity_exact_match": 0.0009900990099009901,
|
| 3656 |
+
"test_entity_fix_rate": 0.2111344208911294,
|
| 3657 |
+
"test_entity_fix_rate_lenient": 0.2051888341543514,
|
| 3658 |
+
"test_entity_fix_rate_recoverable": 0.2082400507902563,
|
| 3659 |
+
"test_entity_hallucination_rate": 0.06732673267326733,
|
| 3660 |
+
"test_entity_preservation_rate": 0.6017693421264275,
|
| 3661 |
+
"test_entity_preservation_rate_lenient": 0.604743209781507,
|
| 3662 |
+
"test_entity_targeted_score": 0.4064518815087785,
|
| 3663 |
+
"test_entity_targeted_score_lenient": 0.4049660219679292,
|
| 3664 |
+
"test_entity_wer": 0.5463007440142612,
|
| 3665 |
+
"test_entity_wer_zwnj_normalized": 0.5412837181498916
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 0.8973237319694013,
|
| 3669 |
+
"step": 5000,
|
| 3670 |
+
"test_typo_exact_match": 0.00558659217877095,
|
| 3671 |
+
"test_typo_fix_rate": 0.2919736331911594,
|
| 3672 |
+
"test_typo_fix_rate_lenient": 0.2889670875595115,
|
| 3673 |
+
"test_typo_fix_rate_recoverable": 0.29298024379991594,
|
| 3674 |
+
"test_typo_hallucination_rate": 0.1005586592178771,
|
| 3675 |
+
"test_typo_preservation_rate": 0.7199894158116015,
|
| 3676 |
+
"test_typo_preservation_rate_lenient": 0.7239004951800522,
|
| 3677 |
+
"test_typo_targeted_score": 0.5059815245013805,
|
| 3678 |
+
"test_typo_targeted_score_lenient": 0.5064337913697818,
|
| 3679 |
+
"test_typo_wer": 0.5167343187562227,
|
| 3680 |
+
"test_typo_wer_zwnj_normalized": 0.5094329989022491
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 0.8973237319694013,
|
| 3684 |
+
"step": 5000,
|
| 3685 |
+
"test_best_wer": 0.2231321104545041
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 0.89911837943334,
|
| 3689 |
+
"grad_norm": 0.7809056043624878,
|
| 3690 |
+
"learning_rate": 5.7961843596524365e-05,
|
| 3691 |
+
"loss": 0.7170194625854492,
|
| 3692 |
+
"step": 5010
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 0.9009130268972788,
|
| 3696 |
+
"grad_norm": 0.957888662815094,
|
| 3697 |
+
"learning_rate": 5.7867397053267855e-05,
|
| 3698 |
+
"loss": 0.6616957664489747,
|
| 3699 |
+
"step": 5020
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 0.9027076743612177,
|
| 3703 |
+
"grad_norm": 1.3222167491912842,
|
| 3704 |
+
"learning_rate": 5.777295051001134e-05,
|
| 3705 |
+
"loss": 0.7434426784515381,
|
| 3706 |
+
"step": 5030
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 0.9045023218251564,
|
| 3710 |
+
"grad_norm": 1.7978832721710205,
|
| 3711 |
+
"learning_rate": 5.7678503966754815e-05,
|
| 3712 |
+
"loss": 0.8404815673828125,
|
| 3713 |
+
"step": 5040
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 0.9062969692890953,
|
| 3717 |
+
"grad_norm": 3.8125853538513184,
|
| 3718 |
+
"learning_rate": 5.7584057423498305e-05,
|
| 3719 |
+
"loss": 0.9580471038818359,
|
| 3720 |
+
"step": 5050
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 0.9080916167530341,
|
| 3724 |
+
"grad_norm": 0.75447016954422,
|
| 3725 |
+
"learning_rate": 5.748961088024178e-05,
|
| 3726 |
+
"loss": 0.6772525310516357,
|
| 3727 |
+
"step": 5060
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 0.9098862642169728,
|
| 3731 |
+
"grad_norm": 1.0550845861434937,
|
| 3732 |
+
"learning_rate": 5.739516433698527e-05,
|
| 3733 |
+
"loss": 0.6701762676239014,
|
| 3734 |
+
"step": 5070
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 0.9116809116809117,
|
| 3738 |
+
"grad_norm": 1.1609597206115723,
|
| 3739 |
+
"learning_rate": 5.7300717793728756e-05,
|
| 3740 |
+
"loss": 0.6679131984710693,
|
| 3741 |
+
"step": 5080
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 0.9134755591448505,
|
| 3745 |
+
"grad_norm": 1.8095602989196777,
|
| 3746 |
+
"learning_rate": 5.720627125047223e-05,
|
| 3747 |
+
"loss": 0.8702205657958985,
|
| 3748 |
+
"step": 5090
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 0.9152702066087893,
|
| 3752 |
+
"grad_norm": 3.851914882659912,
|
| 3753 |
+
"learning_rate": 5.711182470721572e-05,
|
| 3754 |
+
"loss": 0.8425016403198242,
|
| 3755 |
+
"step": 5100
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 0.9170648540727281,
|
| 3759 |
+
"grad_norm": 0.866205096244812,
|
| 3760 |
+
"learning_rate": 5.70173781639592e-05,
|
| 3761 |
+
"loss": 0.6787242412567138,
|
| 3762 |
+
"step": 5110
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 0.9188595015366668,
|
| 3766 |
+
"grad_norm": 0.9801761507987976,
|
| 3767 |
+
"learning_rate": 5.692293162070268e-05,
|
| 3768 |
+
"loss": 0.6703197002410889,
|
| 3769 |
+
"step": 5120
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 0.9206541490006057,
|
| 3773 |
+
"grad_norm": 1.2042592763900757,
|
| 3774 |
+
"learning_rate": 5.682848507744617e-05,
|
| 3775 |
+
"loss": 0.7309328079223633,
|
| 3776 |
+
"step": 5130
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 0.9224487964645445,
|
| 3780 |
+
"grad_norm": 1.6338990926742554,
|
| 3781 |
+
"learning_rate": 5.673403853418965e-05,
|
| 3782 |
+
"loss": 0.8452322959899903,
|
| 3783 |
+
"step": 5140
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 0.9242434439284833,
|
| 3787 |
+
"grad_norm": 4.048831939697266,
|
| 3788 |
+
"learning_rate": 5.663959199093314e-05,
|
| 3789 |
+
"loss": 0.9533859252929687,
|
| 3790 |
+
"step": 5150
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 0.9260380913924221,
|
| 3794 |
+
"grad_norm": 0.8281275629997253,
|
| 3795 |
+
"learning_rate": 5.6545145447676616e-05,
|
| 3796 |
+
"loss": 0.7354491710662842,
|
| 3797 |
+
"step": 5160
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 0.9278327388563609,
|
| 3801 |
+
"grad_norm": 1.0708837509155273,
|
| 3802 |
+
"learning_rate": 5.64506989044201e-05,
|
| 3803 |
+
"loss": 0.7074154853820801,
|
| 3804 |
+
"step": 5170
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 0.9296273863202997,
|
| 3808 |
+
"grad_norm": 1.1584080457687378,
|
| 3809 |
+
"learning_rate": 5.635625236116358e-05,
|
| 3810 |
+
"loss": 0.7574091911315918,
|
| 3811 |
+
"step": 5180
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 0.9314220337842385,
|
| 3815 |
+
"grad_norm": 1.576695203781128,
|
| 3816 |
+
"learning_rate": 5.6261805817907066e-05,
|
| 3817 |
+
"loss": 0.8016712188720703,
|
| 3818 |
+
"step": 5190
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 0.9332166812481774,
|
| 3822 |
+
"grad_norm": 4.139866828918457,
|
| 3823 |
+
"learning_rate": 5.616735927465056e-05,
|
| 3824 |
+
"loss": 0.9117247581481933,
|
| 3825 |
+
"step": 5200
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 0.9350113287121161,
|
| 3829 |
+
"grad_norm": 0.8844179511070251,
|
| 3830 |
+
"learning_rate": 5.607291273139403e-05,
|
| 3831 |
+
"loss": 0.726348876953125,
|
| 3832 |
+
"step": 5210
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 0.9368059761760549,
|
| 3836 |
+
"grad_norm": 0.9993057250976562,
|
| 3837 |
+
"learning_rate": 5.597846618813751e-05,
|
| 3838 |
+
"loss": 0.670358943939209,
|
| 3839 |
+
"step": 5220
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 0.9386006236399937,
|
| 3843 |
+
"grad_norm": 1.256973385810852,
|
| 3844 |
+
"learning_rate": 5.5884019644881e-05,
|
| 3845 |
+
"loss": 0.74342942237854,
|
| 3846 |
+
"step": 5230
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 0.9403952711039325,
|
| 3850 |
+
"grad_norm": 1.7628540992736816,
|
| 3851 |
+
"learning_rate": 5.5789573101624484e-05,
|
| 3852 |
+
"loss": 0.794491958618164,
|
| 3853 |
+
"step": 5240
|
| 3854 |
+
},
|
| 3855 |
+
{
|
| 3856 |
+
"epoch": 0.9421899185678714,
|
| 3857 |
+
"grad_norm": 3.597336530685425,
|
| 3858 |
+
"learning_rate": 5.5695126558367974e-05,
|
| 3859 |
+
"loss": 0.9213722229003907,
|
| 3860 |
+
"step": 5250
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 0.9439845660318101,
|
| 3864 |
+
"grad_norm": 0.9115390777587891,
|
| 3865 |
+
"learning_rate": 5.560068001511145e-05,
|
| 3866 |
+
"loss": 0.6947968482971192,
|
| 3867 |
+
"step": 5260
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 0.9457792134957489,
|
| 3871 |
+
"grad_norm": 1.0560925006866455,
|
| 3872 |
+
"learning_rate": 5.550623347185493e-05,
|
| 3873 |
+
"loss": 0.6915997982025146,
|
| 3874 |
+
"step": 5270
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 0.9475738609596878,
|
| 3878 |
+
"grad_norm": 1.2820700407028198,
|
| 3879 |
+
"learning_rate": 5.541178692859842e-05,
|
| 3880 |
+
"loss": 0.6922568321228028,
|
| 3881 |
+
"step": 5280
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 0.9493685084236265,
|
| 3885 |
+
"grad_norm": 1.6848899126052856,
|
| 3886 |
+
"learning_rate": 5.53173403853419e-05,
|
| 3887 |
+
"loss": 0.8672426223754883,
|
| 3888 |
+
"step": 5290
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 0.9511631558875653,
|
| 3892 |
+
"grad_norm": 4.308657169342041,
|
| 3893 |
+
"learning_rate": 5.522289384208538e-05,
|
| 3894 |
+
"loss": 0.8936351776123047,
|
| 3895 |
+
"step": 5300
|
| 3896 |
+
},
|
| 3897 |
+
{
|
| 3898 |
+
"epoch": 0.9529578033515042,
|
| 3899 |
+
"grad_norm": 0.7795036435127258,
|
| 3900 |
+
"learning_rate": 5.512844729882887e-05,
|
| 3901 |
+
"loss": 0.670732831954956,
|
| 3902 |
+
"step": 5310
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 0.9547524508154429,
|
| 3906 |
+
"grad_norm": 0.9784514307975769,
|
| 3907 |
+
"learning_rate": 5.5034000755572344e-05,
|
| 3908 |
+
"loss": 0.7000267028808593,
|
| 3909 |
+
"step": 5320
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 0.9565470982793818,
|
| 3913 |
+
"grad_norm": 1.1356010437011719,
|
| 3914 |
+
"learning_rate": 5.4939554212315834e-05,
|
| 3915 |
+
"loss": 0.6627930164337158,
|
| 3916 |
+
"step": 5330
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 0.9583417457433205,
|
| 3920 |
+
"grad_norm": 1.6613891124725342,
|
| 3921 |
+
"learning_rate": 5.484510766905932e-05,
|
| 3922 |
+
"loss": 0.804014015197754,
|
| 3923 |
+
"step": 5340
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 0.9601363932072593,
|
| 3927 |
+
"grad_norm": 3.5919458866119385,
|
| 3928 |
+
"learning_rate": 5.4750661125802794e-05,
|
| 3929 |
+
"loss": 0.8332475662231446,
|
| 3930 |
+
"step": 5350
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 0.9619310406711982,
|
| 3934 |
+
"grad_norm": 0.8009278178215027,
|
| 3935 |
+
"learning_rate": 5.4656214582546285e-05,
|
| 3936 |
+
"loss": 0.7539466857910156,
|
| 3937 |
+
"step": 5360
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 0.9637256881351369,
|
| 3941 |
+
"grad_norm": 1.0050593614578247,
|
| 3942 |
+
"learning_rate": 5.456176803928976e-05,
|
| 3943 |
+
"loss": 0.7223537921905517,
|
| 3944 |
+
"step": 5370
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 0.9655203355990758,
|
| 3948 |
+
"grad_norm": 1.1427505016326904,
|
| 3949 |
+
"learning_rate": 5.446732149603325e-05,
|
| 3950 |
+
"loss": 0.6910277843475342,
|
| 3951 |
+
"step": 5380
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 0.9673149830630146,
|
| 3955 |
+
"grad_norm": 1.5899986028671265,
|
| 3956 |
+
"learning_rate": 5.437287495277673e-05,
|
| 3957 |
+
"loss": 0.8054441452026367,
|
| 3958 |
+
"step": 5390
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 0.9691096305269533,
|
| 3962 |
+
"grad_norm": 3.9571502208709717,
|
| 3963 |
+
"learning_rate": 5.427842840952021e-05,
|
| 3964 |
+
"loss": 0.8743718147277832,
|
| 3965 |
+
"step": 5400
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 0.9709042779908922,
|
| 3969 |
+
"grad_norm": 0.8468458652496338,
|
| 3970 |
+
"learning_rate": 5.41839818662637e-05,
|
| 3971 |
+
"loss": 0.6937891006469726,
|
| 3972 |
+
"step": 5410
|
| 3973 |
+
},
|
| 3974 |
+
{
|
| 3975 |
+
"epoch": 0.9726989254548309,
|
| 3976 |
+
"grad_norm": 0.9831876754760742,
|
| 3977 |
+
"learning_rate": 5.408953532300718e-05,
|
| 3978 |
+
"loss": 0.6903666019439697,
|
| 3979 |
+
"step": 5420
|
| 3980 |
+
},
|
| 3981 |
+
{
|
| 3982 |
+
"epoch": 0.9744935729187698,
|
| 3983 |
+
"grad_norm": 1.1358697414398193,
|
| 3984 |
+
"learning_rate": 5.399508877975067e-05,
|
| 3985 |
+
"loss": 0.7664251804351807,
|
| 3986 |
+
"step": 5430
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 0.9762882203827086,
|
| 3990 |
+
"grad_norm": 1.405368685722351,
|
| 3991 |
+
"learning_rate": 5.3900642236494145e-05,
|
| 3992 |
+
"loss": 0.790631103515625,
|
| 3993 |
+
"step": 5440
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 0.9780828678466473,
|
| 3997 |
+
"grad_norm": 4.702672958374023,
|
| 3998 |
+
"learning_rate": 5.380619569323763e-05,
|
| 3999 |
+
"loss": 0.8474712371826172,
|
| 4000 |
+
"step": 5450
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 0.9798775153105862,
|
| 4004 |
+
"grad_norm": 0.8734533190727234,
|
| 4005 |
+
"learning_rate": 5.371174914998112e-05,
|
| 4006 |
+
"loss": 0.7088067054748535,
|
| 4007 |
+
"step": 5460
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 0.981672162774525,
|
| 4011 |
+
"grad_norm": 1.0526103973388672,
|
| 4012 |
+
"learning_rate": 5.3617302606724595e-05,
|
| 4013 |
+
"loss": 0.7186482429504395,
|
| 4014 |
+
"step": 5470
|
| 4015 |
+
},
|
| 4016 |
+
{
|
| 4017 |
+
"epoch": 0.9834668102384638,
|
| 4018 |
+
"grad_norm": 1.2041940689086914,
|
| 4019 |
+
"learning_rate": 5.352285606346807e-05,
|
| 4020 |
+
"loss": 0.6822095394134522,
|
| 4021 |
+
"step": 5480
|
| 4022 |
+
},
|
| 4023 |
+
{
|
| 4024 |
+
"epoch": 0.9852614577024026,
|
| 4025 |
+
"grad_norm": 1.8425958156585693,
|
| 4026 |
+
"learning_rate": 5.342840952021156e-05,
|
| 4027 |
+
"loss": 0.7920569896697998,
|
| 4028 |
+
"step": 5490
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 0.9870561051663413,
|
| 4032 |
+
"grad_norm": 3.488388776779175,
|
| 4033 |
+
"learning_rate": 5.3333962976955046e-05,
|
| 4034 |
+
"loss": 0.8748319625854493,
|
| 4035 |
+
"step": 5500
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 0.9888507526302802,
|
| 4039 |
+
"grad_norm": 0.8285815119743347,
|
| 4040 |
+
"learning_rate": 5.3239516433698536e-05,
|
| 4041 |
+
"loss": 0.7148407936096192,
|
| 4042 |
+
"step": 5510
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 0.990645400094219,
|
| 4046 |
+
"grad_norm": 0.9098016023635864,
|
| 4047 |
+
"learning_rate": 5.314506989044201e-05,
|
| 4048 |
+
"loss": 0.6417829513549804,
|
| 4049 |
+
"step": 5520
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 0.9924400475581578,
|
| 4053 |
+
"grad_norm": 1.148944616317749,
|
| 4054 |
+
"learning_rate": 5.305062334718549e-05,
|
| 4055 |
+
"loss": 0.7329731464385987,
|
| 4056 |
+
"step": 5530
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 0.9942346950220966,
|
| 4060 |
+
"grad_norm": 1.3741636276245117,
|
| 4061 |
+
"learning_rate": 5.295617680392898e-05,
|
| 4062 |
+
"loss": 0.7715582847595215,
|
| 4063 |
+
"step": 5540
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 0.9960293424860354,
|
| 4067 |
+
"grad_norm": 3.777000904083252,
|
| 4068 |
+
"learning_rate": 5.286173026067246e-05,
|
| 4069 |
+
"loss": 0.8748141288757324,
|
| 4070 |
+
"step": 5550
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 0.9978239899499742,
|
| 4074 |
+
"grad_norm": 0.9954040050506592,
|
| 4075 |
+
"learning_rate": 5.2767283717415946e-05,
|
| 4076 |
+
"loss": 0.6957951545715332,
|
| 4077 |
+
"step": 5560
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 0.999618637413913,
|
| 4081 |
+
"grad_norm": 1.751959204673767,
|
| 4082 |
+
"learning_rate": 5.267283717415943e-05,
|
| 4083 |
+
"loss": 0.8281888961791992,
|
| 4084 |
+
"step": 5570
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 1.001256253224757,
|
| 4088 |
+
"grad_norm": 0.7628666162490845,
|
| 4089 |
+
"learning_rate": 5.2578390630902906e-05,
|
| 4090 |
+
"loss": 0.8440546989440918,
|
| 4091 |
+
"step": 5580
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 1.003050900688696,
|
| 4095 |
+
"grad_norm": 1.003116250038147,
|
| 4096 |
+
"learning_rate": 5.2483944087646397e-05,
|
| 4097 |
+
"loss": 0.6586771011352539,
|
| 4098 |
+
"step": 5590
|
| 4099 |
+
},
|
| 4100 |
+
{
|
| 4101 |
+
"epoch": 1.0048455481526348,
|
| 4102 |
+
"grad_norm": 1.100639820098877,
|
| 4103 |
+
"learning_rate": 5.238949754438987e-05,
|
| 4104 |
+
"loss": 0.6179897785186768,
|
| 4105 |
+
"step": 5600
|
| 4106 |
+
},
|
| 4107 |
+
{
|
| 4108 |
+
"epoch": 1.0066401956165736,
|
| 4109 |
+
"grad_norm": 1.2264678478240967,
|
| 4110 |
+
"learning_rate": 5.229505100113336e-05,
|
| 4111 |
+
"loss": 0.6698311328887939,
|
| 4112 |
+
"step": 5610
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 1.0084348430805123,
|
| 4116 |
+
"grad_norm": 2.14436411857605,
|
| 4117 |
+
"learning_rate": 5.220060445787685e-05,
|
| 4118 |
+
"loss": 0.8068962097167969,
|
| 4119 |
+
"step": 5620
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 1.010229490544451,
|
| 4123 |
+
"grad_norm": 0.8558101058006287,
|
| 4124 |
+
"learning_rate": 5.2106157914620323e-05,
|
| 4125 |
+
"loss": 0.7048581600189209,
|
| 4126 |
+
"step": 5630
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 1.01202413800839,
|
| 4130 |
+
"grad_norm": 0.9802150726318359,
|
| 4131 |
+
"learning_rate": 5.2011711371363814e-05,
|
| 4132 |
+
"loss": 0.6395843982696533,
|
| 4133 |
+
"step": 5640
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 1.0138187854723288,
|
| 4137 |
+
"grad_norm": 1.169158935546875,
|
| 4138 |
+
"learning_rate": 5.191726482810729e-05,
|
| 4139 |
+
"loss": 0.5737989902496338,
|
| 4140 |
+
"step": 5650
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 1.0156134329362676,
|
| 4144 |
+
"grad_norm": 1.652338981628418,
|
| 4145 |
+
"learning_rate": 5.1822818284850774e-05,
|
| 4146 |
+
"loss": 0.6900452613830567,
|
| 4147 |
+
"step": 5660
|
| 4148 |
+
},
|
| 4149 |
+
{
|
| 4150 |
+
"epoch": 1.0174080804002064,
|
| 4151 |
+
"grad_norm": 2.533174753189087,
|
| 4152 |
+
"learning_rate": 5.1728371741594264e-05,
|
| 4153 |
+
"loss": 0.7546424388885498,
|
| 4154 |
+
"step": 5670
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 1.0192027278641451,
|
| 4158 |
+
"grad_norm": 0.8599873781204224,
|
| 4159 |
+
"learning_rate": 5.163392519833774e-05,
|
| 4160 |
+
"loss": 0.6629761695861817,
|
| 4161 |
+
"step": 5680
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 1.020997375328084,
|
| 4165 |
+
"grad_norm": 1.0262150764465332,
|
| 4166 |
+
"learning_rate": 5.153947865508123e-05,
|
| 4167 |
+
"loss": 0.6670866966247558,
|
| 4168 |
+
"step": 5690
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 1.0227920227920229,
|
| 4172 |
+
"grad_norm": 1.241133451461792,
|
| 4173 |
+
"learning_rate": 5.144503211182471e-05,
|
| 4174 |
+
"loss": 0.6371388912200928,
|
| 4175 |
+
"step": 5700
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 1.0245866702559616,
|
| 4179 |
+
"grad_norm": 1.4596405029296875,
|
| 4180 |
+
"learning_rate": 5.135058556856819e-05,
|
| 4181 |
+
"loss": 0.6801862716674805,
|
| 4182 |
+
"step": 5710
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 1.0263813177199004,
|
| 4186 |
+
"grad_norm": 2.6000242233276367,
|
| 4187 |
+
"learning_rate": 5.125613902531168e-05,
|
| 4188 |
+
"loss": 0.751060152053833,
|
| 4189 |
+
"step": 5720
|
| 4190 |
+
},
|
| 4191 |
+
{
|
| 4192 |
+
"epoch": 1.0281759651838391,
|
| 4193 |
+
"grad_norm": 0.8851364850997925,
|
| 4194 |
+
"learning_rate": 5.116169248205516e-05,
|
| 4195 |
+
"loss": 0.6396872520446777,
|
| 4196 |
+
"step": 5730
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 1.029970612647778,
|
| 4200 |
+
"grad_norm": 0.9944484829902649,
|
| 4201 |
+
"learning_rate": 5.106724593879865e-05,
|
| 4202 |
+
"loss": 0.6847492218017578,
|
| 4203 |
+
"step": 5740
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 1.0317652601117169,
|
| 4207 |
+
"grad_norm": 1.1320077180862427,
|
| 4208 |
+
"learning_rate": 5.0972799395542125e-05,
|
| 4209 |
+
"loss": 0.5994790077209473,
|
| 4210 |
+
"step": 5750
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 1.0335599075756556,
|
| 4214 |
+
"grad_norm": 1.7546178102493286,
|
| 4215 |
+
"learning_rate": 5.087835285228561e-05,
|
| 4216 |
+
"loss": 0.626808500289917,
|
| 4217 |
+
"step": 5760
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 1.0353545550395944,
|
| 4221 |
+
"grad_norm": 2.7230186462402344,
|
| 4222 |
+
"learning_rate": 5.078390630902909e-05,
|
| 4223 |
+
"loss": 0.6805904865264892,
|
| 4224 |
+
"step": 5770
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 1.0371492025035332,
|
| 4228 |
+
"grad_norm": 0.9365707039833069,
|
| 4229 |
+
"learning_rate": 5.0689459765772575e-05,
|
| 4230 |
+
"loss": 0.7567346096038818,
|
| 4231 |
+
"step": 5780
|
| 4232 |
+
},
|
| 4233 |
+
{
|
| 4234 |
+
"epoch": 1.038943849967472,
|
| 4235 |
+
"grad_norm": 1.0346311330795288,
|
| 4236 |
+
"learning_rate": 5.0595013222516065e-05,
|
| 4237 |
+
"loss": 0.6165975570678711,
|
| 4238 |
+
"step": 5790
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 1.040738497431411,
|
| 4242 |
+
"grad_norm": 1.3088047504425049,
|
| 4243 |
+
"learning_rate": 5.050056667925954e-05,
|
| 4244 |
+
"loss": 0.6644974231719971,
|
| 4245 |
+
"step": 5800
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 1.0425331448953497,
|
| 4249 |
+
"grad_norm": 1.5316091775894165,
|
| 4250 |
+
"learning_rate": 5.040612013600302e-05,
|
| 4251 |
+
"loss": 0.6796547412872315,
|
| 4252 |
+
"step": 5810
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 1.0443277923592884,
|
| 4256 |
+
"grad_norm": 2.469263792037964,
|
| 4257 |
+
"learning_rate": 5.031167359274651e-05,
|
| 4258 |
+
"loss": 0.7114875793457032,
|
| 4259 |
+
"step": 5820
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 1.0461224398232272,
|
| 4263 |
+
"grad_norm": 0.929966926574707,
|
| 4264 |
+
"learning_rate": 5.021722704948999e-05,
|
| 4265 |
+
"loss": 0.7174899101257324,
|
| 4266 |
+
"step": 5830
|
| 4267 |
+
},
|
| 4268 |
+
{
|
| 4269 |
+
"epoch": 1.047917087287166,
|
| 4270 |
+
"grad_norm": 1.0598982572555542,
|
| 4271 |
+
"learning_rate": 5.012278050623347e-05,
|
| 4272 |
+
"loss": 0.6744810581207276,
|
| 4273 |
+
"step": 5840
|
| 4274 |
+
},
|
| 4275 |
+
{
|
| 4276 |
+
"epoch": 1.049711734751105,
|
| 4277 |
+
"grad_norm": 1.2967051267623901,
|
| 4278 |
+
"learning_rate": 5.002833396297696e-05,
|
| 4279 |
+
"loss": 0.6474225044250488,
|
| 4280 |
+
"step": 5850
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 1.0515063822150437,
|
| 4284 |
+
"grad_norm": 1.5724412202835083,
|
| 4285 |
+
"learning_rate": 4.993388741972044e-05,
|
| 4286 |
+
"loss": 0.698382043838501,
|
| 4287 |
+
"step": 5860
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 1.0533010296789824,
|
| 4291 |
+
"grad_norm": 1.998731255531311,
|
| 4292 |
+
"learning_rate": 4.983944087646392e-05,
|
| 4293 |
+
"loss": 0.7015139579772949,
|
| 4294 |
+
"step": 5870
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 1.0550956771429212,
|
| 4298 |
+
"grad_norm": 0.9070185422897339,
|
| 4299 |
+
"learning_rate": 4.974499433320741e-05,
|
| 4300 |
+
"loss": 0.695131492614746,
|
| 4301 |
+
"step": 5880
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 1.05689032460686,
|
| 4305 |
+
"grad_norm": 0.9585561156272888,
|
| 4306 |
+
"learning_rate": 4.965054778995089e-05,
|
| 4307 |
+
"loss": 0.6222951889038086,
|
| 4308 |
+
"step": 5890
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 1.058684972070799,
|
| 4312 |
+
"grad_norm": 1.1435563564300537,
|
| 4313 |
+
"learning_rate": 4.9556101246694376e-05,
|
| 4314 |
+
"loss": 0.5954423904418945,
|
| 4315 |
+
"step": 5900
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 1.0604796195347377,
|
| 4319 |
+
"grad_norm": 1.6986570358276367,
|
| 4320 |
+
"learning_rate": 4.946165470343785e-05,
|
| 4321 |
+
"loss": 0.7150691032409668,
|
| 4322 |
+
"step": 5910
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 1.0622742669986764,
|
| 4326 |
+
"grad_norm": 2.584141492843628,
|
| 4327 |
+
"learning_rate": 4.9367208160181336e-05,
|
| 4328 |
+
"loss": 0.65278639793396,
|
| 4329 |
+
"step": 5920
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 1.0640689144626152,
|
| 4333 |
+
"grad_norm": 0.8705131411552429,
|
| 4334 |
+
"learning_rate": 4.9272761616924826e-05,
|
| 4335 |
+
"loss": 0.6868424415588379,
|
| 4336 |
+
"step": 5930
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 1.065863561926554,
|
| 4340 |
+
"grad_norm": 1.0080828666687012,
|
| 4341 |
+
"learning_rate": 4.917831507366831e-05,
|
| 4342 |
+
"loss": 0.6452962398529053,
|
| 4343 |
+
"step": 5940
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 1.067658209390493,
|
| 4347 |
+
"grad_norm": 1.149987816810608,
|
| 4348 |
+
"learning_rate": 4.908386853041179e-05,
|
| 4349 |
+
"loss": 0.6242999076843262,
|
| 4350 |
+
"step": 5950
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 1.0694528568544317,
|
| 4354 |
+
"grad_norm": 1.478263258934021,
|
| 4355 |
+
"learning_rate": 4.898942198715527e-05,
|
| 4356 |
+
"loss": 0.6609876155853271,
|
| 4357 |
+
"step": 5960
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 1.0712475043183705,
|
| 4361 |
+
"grad_norm": 2.334758758544922,
|
| 4362 |
+
"learning_rate": 4.889497544389875e-05,
|
| 4363 |
+
"loss": 0.7123387813568115,
|
| 4364 |
+
"step": 5970
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 1.0730421517823092,
|
| 4368 |
+
"grad_norm": 0.8998429775238037,
|
| 4369 |
+
"learning_rate": 4.8800528900642236e-05,
|
| 4370 |
+
"loss": 0.6774604320526123,
|
| 4371 |
+
"step": 5980
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 1.074836799246248,
|
| 4375 |
+
"grad_norm": 1.1068638563156128,
|
| 4376 |
+
"learning_rate": 4.870608235738573e-05,
|
| 4377 |
+
"loss": 0.6981619834899903,
|
| 4378 |
+
"step": 5990
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 1.076631446710187,
|
| 4382 |
+
"grad_norm": 1.2205283641815186,
|
| 4383 |
+
"learning_rate": 4.86116358141292e-05,
|
| 4384 |
+
"loss": 0.6537732124328614,
|
| 4385 |
+
"step": 6000
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 1.076631446710187,
|
| 4389 |
+
"step": 6000,
|
| 4390 |
+
"test_entity_exact_match": 0.0049504950495049506,
|
| 4391 |
+
"test_entity_fix_rate": 0.23884522653975013,
|
| 4392 |
+
"test_entity_fix_rate_lenient": 0.2317241379310345,
|
| 4393 |
+
"test_entity_fix_rate_recoverable": 0.23510542319644467,
|
| 4394 |
+
"test_entity_hallucination_rate": 0.005940594059405941,
|
| 4395 |
+
"test_entity_preservation_rate": 0.6887630690043429,
|
| 4396 |
+
"test_entity_preservation_rate_lenient": 0.6904019604069641,
|
| 4397 |
+
"test_entity_targeted_score": 0.46380414777204654,
|
| 4398 |
+
"test_entity_targeted_score_lenient": 0.4610630491689993,
|
| 4399 |
+
"test_entity_wer": 0.40874883914561116,
|
| 4400 |
+
"test_entity_wer_zwnj_normalized": 0.40443099454531867
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 1.076631446710187,
|
| 4404 |
+
"step": 6000,
|
| 4405 |
+
"test_typo_exact_match": 0.00558659217877095,
|
| 4406 |
+
"test_typo_fix_rate": 0.3352074447460256,
|
| 4407 |
+
"test_typo_fix_rate_lenient": 0.327882425998758,
|
| 4408 |
+
"test_typo_fix_rate_recoverable": 0.3322824716267339,
|
| 4409 |
+
"test_typo_hallucination_rate": 0.01675977653631285,
|
| 4410 |
+
"test_typo_preservation_rate": 0.8174521506482816,
|
| 4411 |
+
"test_typo_preservation_rate_lenient": 0.8203580532684813,
|
| 4412 |
+
"test_typo_targeted_score": 0.5763297976971535,
|
| 4413 |
+
"test_typo_targeted_score_lenient": 0.5741202396336197,
|
| 4414 |
+
"test_typo_wer": 0.2884021342319573,
|
| 4415 |
+
"test_typo_wer_zwnj_normalized": 0.2824538561691047
|
| 4416 |
+
},
|
| 4417 |
+
{
|
| 4418 |
+
"epoch": 1.076631446710187,
|
| 4419 |
+
"step": 6000,
|
| 4420 |
+
"test_best_wer": 0.2231321104545041
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 1.0784260941741257,
|
| 4424 |
+
"grad_norm": 1.5345807075500488,
|
| 4425 |
+
"learning_rate": 4.851718927087269e-05,
|
| 4426 |
+
"loss": 0.6438473701477051,
|
| 4427 |
+
"step": 6010
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 1.0802207416380645,
|
| 4431 |
+
"grad_norm": 2.6961584091186523,
|
| 4432 |
+
"learning_rate": 4.842274272761617e-05,
|
| 4433 |
+
"loss": 0.7586267471313477,
|
| 4434 |
+
"step": 6020
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 1.0820153891020032,
|
| 4438 |
+
"grad_norm": 0.9163028001785278,
|
| 4439 |
+
"learning_rate": 4.8328296184359654e-05,
|
| 4440 |
+
"loss": 0.7256781101226807,
|
| 4441 |
+
"step": 6030
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 1.083810036565942,
|
| 4445 |
+
"grad_norm": 1.1057441234588623,
|
| 4446 |
+
"learning_rate": 4.8233849641103144e-05,
|
| 4447 |
+
"loss": 0.5618266582489013,
|
| 4448 |
+
"step": 6040
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 1.085604684029881,
|
| 4452 |
+
"grad_norm": 1.2693841457366943,
|
| 4453 |
+
"learning_rate": 4.813940309784662e-05,
|
| 4454 |
+
"loss": 0.6748818397521973,
|
| 4455 |
+
"step": 6050
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 1.0873993314938197,
|
| 4459 |
+
"grad_norm": 1.6324713230133057,
|
| 4460 |
+
"learning_rate": 4.8044956554590104e-05,
|
| 4461 |
+
"loss": 0.6527456283569336,
|
| 4462 |
+
"step": 6060
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 1.0891939789577585,
|
| 4466 |
+
"grad_norm": 2.866149663925171,
|
| 4467 |
+
"learning_rate": 4.795051001133359e-05,
|
| 4468 |
+
"loss": 0.7118582725524902,
|
| 4469 |
+
"step": 6070
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 1.0909886264216972,
|
| 4473 |
+
"grad_norm": 0.8941351175308228,
|
| 4474 |
+
"learning_rate": 4.785606346807707e-05,
|
| 4475 |
+
"loss": 0.6796853065490722,
|
| 4476 |
+
"step": 6080
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 1.092783273885636,
|
| 4480 |
+
"grad_norm": 0.9570082426071167,
|
| 4481 |
+
"learning_rate": 4.7761616924820554e-05,
|
| 4482 |
+
"loss": 0.6397138118743897,
|
| 4483 |
+
"step": 6090
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 1.094577921349575,
|
| 4487 |
+
"grad_norm": 1.2090080976486206,
|
| 4488 |
+
"learning_rate": 4.766717038156404e-05,
|
| 4489 |
+
"loss": 0.6042757987976074,
|
| 4490 |
+
"step": 6100
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 1.0963725688135137,
|
| 4494 |
+
"grad_norm": 1.5790772438049316,
|
| 4495 |
+
"learning_rate": 4.757272383830752e-05,
|
| 4496 |
+
"loss": 0.6695082187652588,
|
| 4497 |
+
"step": 6110
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 1.0981672162774525,
|
| 4501 |
+
"grad_norm": 2.5716567039489746,
|
| 4502 |
+
"learning_rate": 4.7478277295051004e-05,
|
| 4503 |
+
"loss": 0.7349279403686524,
|
| 4504 |
+
"step": 6120
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 1.0999618637413913,
|
| 4508 |
+
"grad_norm": 0.8835099935531616,
|
| 4509 |
+
"learning_rate": 4.738383075179449e-05,
|
| 4510 |
+
"loss": 0.6819751739501954,
|
| 4511 |
+
"step": 6130
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 1.10175651120533,
|
| 4515 |
+
"grad_norm": 1.1461682319641113,
|
| 4516 |
+
"learning_rate": 4.728938420853797e-05,
|
| 4517 |
+
"loss": 0.6171199798583984,
|
| 4518 |
+
"step": 6140
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 1.103551158669269,
|
| 4522 |
+
"grad_norm": 1.4509607553482056,
|
| 4523 |
+
"learning_rate": 4.7194937665281455e-05,
|
| 4524 |
+
"loss": 0.628743028640747,
|
| 4525 |
+
"step": 6150
|
| 4526 |
+
},
|
| 4527 |
+
{
|
| 4528 |
+
"epoch": 1.1053458061332078,
|
| 4529 |
+
"grad_norm": 1.5972394943237305,
|
| 4530 |
+
"learning_rate": 4.710049112202494e-05,
|
| 4531 |
+
"loss": 0.6271132946014404,
|
| 4532 |
+
"step": 6160
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 1.1071404535971465,
|
| 4536 |
+
"grad_norm": 2.8629398345947266,
|
| 4537 |
+
"learning_rate": 4.700604457876842e-05,
|
| 4538 |
+
"loss": 0.7839233875274658,
|
| 4539 |
+
"step": 6170
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 1.1089351010610853,
|
| 4543 |
+
"grad_norm": 0.8886165618896484,
|
| 4544 |
+
"learning_rate": 4.69115980355119e-05,
|
| 4545 |
+
"loss": 0.6798130989074707,
|
| 4546 |
+
"step": 6180
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 1.110729748525024,
|
| 4550 |
+
"grad_norm": 1.0677049160003662,
|
| 4551 |
+
"learning_rate": 4.681715149225538e-05,
|
| 4552 |
+
"loss": 0.5951868057250976,
|
| 4553 |
+
"step": 6190
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 1.112524395988963,
|
| 4557 |
+
"grad_norm": 1.3017163276672363,
|
| 4558 |
+
"learning_rate": 4.672270494899887e-05,
|
| 4559 |
+
"loss": 0.6186211109161377,
|
| 4560 |
+
"step": 6200
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 1.1143190434529018,
|
| 4564 |
+
"grad_norm": 1.8037720918655396,
|
| 4565 |
+
"learning_rate": 4.6628258405742355e-05,
|
| 4566 |
+
"loss": 0.6517776489257813,
|
| 4567 |
+
"step": 6210
|
| 4568 |
+
},
|
| 4569 |
+
{
|
| 4570 |
+
"epoch": 1.1161136909168405,
|
| 4571 |
+
"grad_norm": 2.990157127380371,
|
| 4572 |
+
"learning_rate": 4.653381186248584e-05,
|
| 4573 |
+
"loss": 0.7219930171966553,
|
| 4574 |
+
"step": 6220
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 1.1179083383807793,
|
| 4578 |
+
"grad_norm": 0.865894079208374,
|
| 4579 |
+
"learning_rate": 4.6439365319229315e-05,
|
| 4580 |
+
"loss": 0.6599620819091797,
|
| 4581 |
+
"step": 6230
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 1.119702985844718,
|
| 4585 |
+
"grad_norm": 1.0573729276657104,
|
| 4586 |
+
"learning_rate": 4.63449187759728e-05,
|
| 4587 |
+
"loss": 0.595738697052002,
|
| 4588 |
+
"step": 6240
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 1.1214976333086568,
|
| 4592 |
+
"grad_norm": 1.2556575536727905,
|
| 4593 |
+
"learning_rate": 4.625047223271628e-05,
|
| 4594 |
+
"loss": 0.5770184516906738,
|
| 4595 |
+
"step": 6250
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 1.1232922807725958,
|
| 4599 |
+
"grad_norm": 1.5537583827972412,
|
| 4600 |
+
"learning_rate": 4.615602568945977e-05,
|
| 4601 |
+
"loss": 0.6816306114196777,
|
| 4602 |
+
"step": 6260
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 1.1250869282365346,
|
| 4606 |
+
"grad_norm": 2.8041980266571045,
|
| 4607 |
+
"learning_rate": 4.606157914620325e-05,
|
| 4608 |
+
"loss": 0.7952973365783691,
|
| 4609 |
+
"step": 6270
|
| 4610 |
+
},
|
| 4611 |
+
{
|
| 4612 |
+
"epoch": 1.1268815757004733,
|
| 4613 |
+
"grad_norm": 0.9291264414787292,
|
| 4614 |
+
"learning_rate": 4.596713260294673e-05,
|
| 4615 |
+
"loss": 0.7760135173797608,
|
| 4616 |
+
"step": 6280
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 1.128676223164412,
|
| 4620 |
+
"grad_norm": 1.136435866355896,
|
| 4621 |
+
"learning_rate": 4.5872686059690216e-05,
|
| 4622 |
+
"loss": 0.6364834785461426,
|
| 4623 |
+
"step": 6290
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 1.130470870628351,
|
| 4627 |
+
"grad_norm": 1.1664420366287231,
|
| 4628 |
+
"learning_rate": 4.57782395164337e-05,
|
| 4629 |
+
"loss": 0.6077825546264648,
|
| 4630 |
+
"step": 6300
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 1.1322655180922898,
|
| 4634 |
+
"grad_norm": 1.5633567571640015,
|
| 4635 |
+
"learning_rate": 4.568379297317719e-05,
|
| 4636 |
+
"loss": 0.7070342063903808,
|
| 4637 |
+
"step": 6310
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 1.1340601655562286,
|
| 4641 |
+
"grad_norm": 2.517033815383911,
|
| 4642 |
+
"learning_rate": 4.5589346429920666e-05,
|
| 4643 |
+
"loss": 0.7303966045379638,
|
| 4644 |
+
"step": 6320
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 1.1358548130201673,
|
| 4648 |
+
"grad_norm": 0.9475764632225037,
|
| 4649 |
+
"learning_rate": 4.549489988666415e-05,
|
| 4650 |
+
"loss": 0.7106866359710693,
|
| 4651 |
+
"step": 6330
|
| 4652 |
+
},
|
| 4653 |
+
{
|
| 4654 |
+
"epoch": 1.137649460484106,
|
| 4655 |
+
"grad_norm": 1.0835204124450684,
|
| 4656 |
+
"learning_rate": 4.540045334340763e-05,
|
| 4657 |
+
"loss": 0.6016337394714355,
|
| 4658 |
+
"step": 6340
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 1.1394441079480448,
|
| 4662 |
+
"grad_norm": 1.4215147495269775,
|
| 4663 |
+
"learning_rate": 4.5306006800151116e-05,
|
| 4664 |
+
"loss": 0.6349590301513672,
|
| 4665 |
+
"step": 6350
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 1.1412387554119838,
|
| 4669 |
+
"grad_norm": 1.699916958808899,
|
| 4670 |
+
"learning_rate": 4.52115602568946e-05,
|
| 4671 |
+
"loss": 0.6965232849121094,
|
| 4672 |
+
"step": 6360
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 1.1430334028759226,
|
| 4676 |
+
"grad_norm": 2.6666152477264404,
|
| 4677 |
+
"learning_rate": 4.511711371363808e-05,
|
| 4678 |
+
"loss": 0.7466271400451661,
|
| 4679 |
+
"step": 6370
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 1.1448280503398613,
|
| 4683 |
+
"grad_norm": 0.8652181029319763,
|
| 4684 |
+
"learning_rate": 4.5022667170381567e-05,
|
| 4685 |
+
"loss": 0.7699360370635986,
|
| 4686 |
+
"step": 6380
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 1.1466226978038,
|
| 4690 |
+
"grad_norm": 1.0938860177993774,
|
| 4691 |
+
"learning_rate": 4.492822062712505e-05,
|
| 4692 |
+
"loss": 0.6609859943389893,
|
| 4693 |
+
"step": 6390
|
| 4694 |
+
},
|
| 4695 |
+
{
|
| 4696 |
+
"epoch": 1.148417345267739,
|
| 4697 |
+
"grad_norm": 1.3298280239105225,
|
| 4698 |
+
"learning_rate": 4.483377408386853e-05,
|
| 4699 |
+
"loss": 0.6199182510375977,
|
| 4700 |
+
"step": 6400
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 1.1502119927316778,
|
| 4704 |
+
"grad_norm": 1.7540221214294434,
|
| 4705 |
+
"learning_rate": 4.473932754061202e-05,
|
| 4706 |
+
"loss": 0.67717604637146,
|
| 4707 |
+
"step": 6410
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 1.1520066401956166,
|
| 4711 |
+
"grad_norm": 2.5371453762054443,
|
| 4712 |
+
"learning_rate": 4.46448809973555e-05,
|
| 4713 |
+
"loss": 0.793154239654541,
|
| 4714 |
+
"step": 6420
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 1.1538012876595554,
|
| 4718 |
+
"grad_norm": 0.9699407815933228,
|
| 4719 |
+
"learning_rate": 4.4550434454098984e-05,
|
| 4720 |
+
"loss": 0.6550703525543213,
|
| 4721 |
+
"step": 6430
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 1.1555959351234941,
|
| 4725 |
+
"grad_norm": 1.0454729795455933,
|
| 4726 |
+
"learning_rate": 4.445598791084247e-05,
|
| 4727 |
+
"loss": 0.6019119262695313,
|
| 4728 |
+
"step": 6440
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 1.1573905825874329,
|
| 4732 |
+
"grad_norm": 1.233948826789856,
|
| 4733 |
+
"learning_rate": 4.4361541367585944e-05,
|
| 4734 |
+
"loss": 0.6351359367370606,
|
| 4735 |
+
"step": 6450
|
| 4736 |
+
},
|
| 4737 |
+
{
|
| 4738 |
+
"epoch": 1.1591852300513719,
|
| 4739 |
+
"grad_norm": 1.5669020414352417,
|
| 4740 |
+
"learning_rate": 4.426709482432943e-05,
|
| 4741 |
+
"loss": 0.6300336360931397,
|
| 4742 |
+
"step": 6460
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 1.1609798775153106,
|
| 4746 |
+
"grad_norm": 2.9245171546936035,
|
| 4747 |
+
"learning_rate": 4.417264828107292e-05,
|
| 4748 |
+
"loss": 0.772235631942749,
|
| 4749 |
+
"step": 6470
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 1.1627745249792494,
|
| 4753 |
+
"grad_norm": 0.8748898506164551,
|
| 4754 |
+
"learning_rate": 4.40782017378164e-05,
|
| 4755 |
+
"loss": 0.6726730823516845,
|
| 4756 |
+
"step": 6480
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 1.1645691724431881,
|
| 4760 |
+
"grad_norm": 1.2944382429122925,
|
| 4761 |
+
"learning_rate": 4.3983755194559884e-05,
|
| 4762 |
+
"loss": 0.6441913604736328,
|
| 4763 |
+
"step": 6490
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 1.166363819907127,
|
| 4767 |
+
"grad_norm": 1.2455261945724487,
|
| 4768 |
+
"learning_rate": 4.388930865130336e-05,
|
| 4769 |
+
"loss": 0.6275741577148437,
|
| 4770 |
+
"step": 6500
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 1.1681584673710659,
|
| 4774 |
+
"grad_norm": 1.5028836727142334,
|
| 4775 |
+
"learning_rate": 4.3794862108046844e-05,
|
| 4776 |
+
"loss": 0.7088084697723389,
|
| 4777 |
+
"step": 6510
|
| 4778 |
+
},
|
| 4779 |
+
{
|
| 4780 |
+
"epoch": 1.1699531148350046,
|
| 4781 |
+
"grad_norm": 2.693411350250244,
|
| 4782 |
+
"learning_rate": 4.3700415564790334e-05,
|
| 4783 |
+
"loss": 0.7553691387176513,
|
| 4784 |
+
"step": 6520
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 1.1717477622989434,
|
| 4788 |
+
"grad_norm": 0.8517025709152222,
|
| 4789 |
+
"learning_rate": 4.360596902153382e-05,
|
| 4790 |
+
"loss": 0.6350733280181885,
|
| 4791 |
+
"step": 6530
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 1.1735424097628822,
|
| 4795 |
+
"grad_norm": 1.1345810890197754,
|
| 4796 |
+
"learning_rate": 4.3511522478277295e-05,
|
| 4797 |
+
"loss": 0.6504807949066163,
|
| 4798 |
+
"step": 6540
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 1.175337057226821,
|
| 4802 |
+
"grad_norm": 1.212336778640747,
|
| 4803 |
+
"learning_rate": 4.341707593502078e-05,
|
| 4804 |
+
"loss": 0.5438730716705322,
|
| 4805 |
+
"step": 6550
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 1.17713170469076,
|
| 4809 |
+
"grad_norm": 1.5869418382644653,
|
| 4810 |
+
"learning_rate": 4.332262939176426e-05,
|
| 4811 |
+
"loss": 0.6788577079772949,
|
| 4812 |
+
"step": 6560
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 1.1789263521546987,
|
| 4816 |
+
"grad_norm": 2.5476667881011963,
|
| 4817 |
+
"learning_rate": 4.3228182848507745e-05,
|
| 4818 |
+
"loss": 0.7301935672760009,
|
| 4819 |
+
"step": 6570
|
| 4820 |
+
},
|
| 4821 |
+
{
|
| 4822 |
+
"epoch": 1.1807209996186374,
|
| 4823 |
+
"grad_norm": 0.8157092332839966,
|
| 4824 |
+
"learning_rate": 4.3133736305251235e-05,
|
| 4825 |
+
"loss": 0.7241593360900879,
|
| 4826 |
+
"step": 6580
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 1.1825156470825762,
|
| 4830 |
+
"grad_norm": 1.0789809226989746,
|
| 4831 |
+
"learning_rate": 4.303928976199471e-05,
|
| 4832 |
+
"loss": 0.5883336544036866,
|
| 4833 |
+
"step": 6590
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 1.184310294546515,
|
| 4837 |
+
"grad_norm": 1.346107840538025,
|
| 4838 |
+
"learning_rate": 4.2944843218738195e-05,
|
| 4839 |
+
"loss": 0.5695505619049073,
|
| 4840 |
+
"step": 6600
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 1.186104942010454,
|
| 4844 |
+
"grad_norm": 1.7241944074630737,
|
| 4845 |
+
"learning_rate": 4.285039667548168e-05,
|
| 4846 |
+
"loss": 0.6097249984741211,
|
| 4847 |
+
"step": 6610
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 1.1878995894743927,
|
| 4851 |
+
"grad_norm": 2.834289789199829,
|
| 4852 |
+
"learning_rate": 4.275595013222516e-05,
|
| 4853 |
+
"loss": 0.7102833271026612,
|
| 4854 |
+
"step": 6620
|
| 4855 |
+
},
|
| 4856 |
+
{
|
| 4857 |
+
"epoch": 1.1896942369383314,
|
| 4858 |
+
"grad_norm": 0.8784908056259155,
|
| 4859 |
+
"learning_rate": 4.2661503588968645e-05,
|
| 4860 |
+
"loss": 0.6898719310760498,
|
| 4861 |
+
"step": 6630
|
| 4862 |
+
},
|
| 4863 |
+
{
|
| 4864 |
+
"epoch": 1.1914888844022702,
|
| 4865 |
+
"grad_norm": 1.1728085279464722,
|
| 4866 |
+
"learning_rate": 4.256705704571213e-05,
|
| 4867 |
+
"loss": 0.5820427417755127,
|
| 4868 |
+
"step": 6640
|
| 4869 |
+
},
|
| 4870 |
+
{
|
| 4871 |
+
"epoch": 1.193283531866209,
|
| 4872 |
+
"grad_norm": 1.3177632093429565,
|
| 4873 |
+
"learning_rate": 4.247261050245561e-05,
|
| 4874 |
+
"loss": 0.6320508480072021,
|
| 4875 |
+
"step": 6650
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 1.195078179330148,
|
| 4879 |
+
"grad_norm": 1.6765938997268677,
|
| 4880 |
+
"learning_rate": 4.2378163959199096e-05,
|
| 4881 |
+
"loss": 0.6680030345916748,
|
| 4882 |
+
"step": 6660
|
| 4883 |
+
},
|
| 4884 |
+
{
|
| 4885 |
+
"epoch": 1.1968728267940867,
|
| 4886 |
+
"grad_norm": 2.4049787521362305,
|
| 4887 |
+
"learning_rate": 4.228371741594258e-05,
|
| 4888 |
+
"loss": 0.6540277481079102,
|
| 4889 |
+
"step": 6670
|
| 4890 |
+
},
|
| 4891 |
+
{
|
| 4892 |
+
"epoch": 1.1986674742580254,
|
| 4893 |
+
"grad_norm": 0.8671777844429016,
|
| 4894 |
+
"learning_rate": 4.218927087268606e-05,
|
| 4895 |
+
"loss": 0.6770136833190918,
|
| 4896 |
+
"step": 6680
|
| 4897 |
+
},
|
| 4898 |
+
{
|
| 4899 |
+
"epoch": 1.2004621217219642,
|
| 4900 |
+
"grad_norm": 0.9750893115997314,
|
| 4901 |
+
"learning_rate": 4.2094824329429546e-05,
|
| 4902 |
+
"loss": 0.6304139137268067,
|
| 4903 |
+
"step": 6690
|
| 4904 |
+
},
|
| 4905 |
+
{
|
| 4906 |
+
"epoch": 1.202256769185903,
|
| 4907 |
+
"grad_norm": 1.142932653427124,
|
| 4908 |
+
"learning_rate": 4.200037778617303e-05,
|
| 4909 |
+
"loss": 0.5581915378570557,
|
| 4910 |
+
"step": 6700
|
| 4911 |
+
},
|
| 4912 |
+
{
|
| 4913 |
+
"epoch": 1.204051416649842,
|
| 4914 |
+
"grad_norm": 1.5189772844314575,
|
| 4915 |
+
"learning_rate": 4.190593124291651e-05,
|
| 4916 |
+
"loss": 0.6183357238769531,
|
| 4917 |
+
"step": 6710
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 1.2058460641137807,
|
| 4921 |
+
"grad_norm": 2.766662836074829,
|
| 4922 |
+
"learning_rate": 4.181148469965999e-05,
|
| 4923 |
+
"loss": 0.689277458190918,
|
| 4924 |
+
"step": 6720
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 1.2076407115777195,
|
| 4928 |
+
"grad_norm": 0.8908812403678894,
|
| 4929 |
+
"learning_rate": 4.171703815640348e-05,
|
| 4930 |
+
"loss": 0.7036964893341064,
|
| 4931 |
+
"step": 6730
|
| 4932 |
+
},
|
| 4933 |
+
{
|
| 4934 |
+
"epoch": 1.2094353590416582,
|
| 4935 |
+
"grad_norm": 1.0966931581497192,
|
| 4936 |
+
"learning_rate": 4.162259161314696e-05,
|
| 4937 |
+
"loss": 0.5891323089599609,
|
| 4938 |
+
"step": 6740
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 1.211230006505597,
|
| 4942 |
+
"grad_norm": 1.1278657913208008,
|
| 4943 |
+
"learning_rate": 4.1528145069890446e-05,
|
| 4944 |
+
"loss": 0.6117199420928955,
|
| 4945 |
+
"step": 6750
|
| 4946 |
+
},
|
| 4947 |
+
{
|
| 4948 |
+
"epoch": 1.2130246539695357,
|
| 4949 |
+
"grad_norm": 1.713985562324524,
|
| 4950 |
+
"learning_rate": 4.143369852663393e-05,
|
| 4951 |
+
"loss": 0.6227171421051025,
|
| 4952 |
+
"step": 6760
|
| 4953 |
+
},
|
| 4954 |
+
{
|
| 4955 |
+
"epoch": 1.2148193014334747,
|
| 4956 |
+
"grad_norm": 3.1219239234924316,
|
| 4957 |
+
"learning_rate": 4.1339251983377406e-05,
|
| 4958 |
+
"loss": 0.7286433219909668,
|
| 4959 |
+
"step": 6770
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 1.2166139488974135,
|
| 4963 |
+
"grad_norm": 0.9535353779792786,
|
| 4964 |
+
"learning_rate": 4.124480544012089e-05,
|
| 4965 |
+
"loss": 0.6415516853332519,
|
| 4966 |
+
"step": 6780
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 1.2184085963613522,
|
| 4970 |
+
"grad_norm": 1.073463797569275,
|
| 4971 |
+
"learning_rate": 4.115035889686438e-05,
|
| 4972 |
+
"loss": 0.6185556411743164,
|
| 4973 |
+
"step": 6790
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 1.220203243825291,
|
| 4977 |
+
"grad_norm": 1.289807915687561,
|
| 4978 |
+
"learning_rate": 4.1055912353607863e-05,
|
| 4979 |
+
"loss": 0.5423048496246338,
|
| 4980 |
+
"step": 6800
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 1.22199789128923,
|
| 4984 |
+
"grad_norm": 1.679877519607544,
|
| 4985 |
+
"learning_rate": 4.096146581035134e-05,
|
| 4986 |
+
"loss": 0.6980320930480957,
|
| 4987 |
+
"step": 6810
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 1.2237925387531687,
|
| 4991 |
+
"grad_norm": 2.62693190574646,
|
| 4992 |
+
"learning_rate": 4.0867019267094824e-05,
|
| 4993 |
+
"loss": 0.7437369346618652,
|
| 4994 |
+
"step": 6820
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 1.2255871862171075,
|
| 4998 |
+
"grad_norm": 0.9065632820129395,
|
| 4999 |
+
"learning_rate": 4.077257272383831e-05,
|
| 5000 |
+
"loss": 0.6568538188934326,
|
| 5001 |
+
"step": 6830
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 1.2273818336810463,
|
| 5005 |
+
"grad_norm": 1.1957964897155762,
|
| 5006 |
+
"learning_rate": 4.067812618058179e-05,
|
| 5007 |
+
"loss": 0.6349117755889893,
|
| 5008 |
+
"step": 6840
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 1.229176481144985,
|
| 5012 |
+
"grad_norm": 1.3036094903945923,
|
| 5013 |
+
"learning_rate": 4.058367963732528e-05,
|
| 5014 |
+
"loss": 0.5703445911407471,
|
| 5015 |
+
"step": 6850
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 1.2309711286089238,
|
| 5019 |
+
"grad_norm": 1.6722497940063477,
|
| 5020 |
+
"learning_rate": 4.048923309406876e-05,
|
| 5021 |
+
"loss": 0.6664234161376953,
|
| 5022 |
+
"step": 6860
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 1.2327657760728628,
|
| 5026 |
+
"grad_norm": 2.804495096206665,
|
| 5027 |
+
"learning_rate": 4.039478655081224e-05,
|
| 5028 |
+
"loss": 0.7480207920074463,
|
| 5029 |
+
"step": 6870
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 1.2345604235368015,
|
| 5033 |
+
"grad_norm": 0.8401798605918884,
|
| 5034 |
+
"learning_rate": 4.0300340007555724e-05,
|
| 5035 |
+
"loss": 0.6777836322784424,
|
| 5036 |
+
"step": 6880
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 1.2363550710007403,
|
| 5040 |
+
"grad_norm": 1.1064029932022095,
|
| 5041 |
+
"learning_rate": 4.020589346429921e-05,
|
| 5042 |
+
"loss": 0.6124742507934571,
|
| 5043 |
+
"step": 6890
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 1.238149718464679,
|
| 5047 |
+
"grad_norm": 1.3008829355239868,
|
| 5048 |
+
"learning_rate": 4.011144692104269e-05,
|
| 5049 |
+
"loss": 0.6095581531524659,
|
| 5050 |
+
"step": 6900
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 1.239944365928618,
|
| 5054 |
+
"grad_norm": 1.7268178462982178,
|
| 5055 |
+
"learning_rate": 4.0017000377786174e-05,
|
| 5056 |
+
"loss": 0.665320348739624,
|
| 5057 |
+
"step": 6910
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 1.2417390133925568,
|
| 5061 |
+
"grad_norm": 3.6321637630462646,
|
| 5062 |
+
"learning_rate": 3.992255383452966e-05,
|
| 5063 |
+
"loss": 0.736395263671875,
|
| 5064 |
+
"step": 6920
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 1.2435336608564955,
|
| 5068 |
+
"grad_norm": 0.8264207243919373,
|
| 5069 |
+
"learning_rate": 3.982810729127314e-05,
|
| 5070 |
+
"loss": 0.7000748634338378,
|
| 5071 |
+
"step": 6930
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 1.2453283083204343,
|
| 5075 |
+
"grad_norm": 1.0241384506225586,
|
| 5076 |
+
"learning_rate": 3.9733660748016625e-05,
|
| 5077 |
+
"loss": 0.6075915813446044,
|
| 5078 |
+
"step": 6940
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 1.247122955784373,
|
| 5082 |
+
"grad_norm": 1.3497692346572876,
|
| 5083 |
+
"learning_rate": 3.963921420476011e-05,
|
| 5084 |
+
"loss": 0.5755562305450439,
|
| 5085 |
+
"step": 6950
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 1.2489176032483118,
|
| 5089 |
+
"grad_norm": 1.7268356084823608,
|
| 5090 |
+
"learning_rate": 3.954476766150359e-05,
|
| 5091 |
+
"loss": 0.6569338798522949,
|
| 5092 |
+
"step": 6960
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 1.2507122507122508,
|
| 5096 |
+
"grad_norm": 2.9455223083496094,
|
| 5097 |
+
"learning_rate": 3.9450321118247075e-05,
|
| 5098 |
+
"loss": 0.6878787517547608,
|
| 5099 |
+
"step": 6970
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 1.2525068981761895,
|
| 5103 |
+
"grad_norm": 0.8486849069595337,
|
| 5104 |
+
"learning_rate": 3.935587457499056e-05,
|
| 5105 |
+
"loss": 0.6220896720886231,
|
| 5106 |
+
"step": 6980
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 1.2543015456401283,
|
| 5110 |
+
"grad_norm": 1.1171101331710815,
|
| 5111 |
+
"learning_rate": 3.9261428031734035e-05,
|
| 5112 |
+
"loss": 0.579223918914795,
|
| 5113 |
+
"step": 6990
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 1.256096193104067,
|
| 5117 |
+
"grad_norm": 1.3631322383880615,
|
| 5118 |
+
"learning_rate": 3.9166981488477525e-05,
|
| 5119 |
+
"loss": 0.6380829811096191,
|
| 5120 |
+
"step": 7000
|
| 5121 |
}
|
| 5122 |
],
|
| 5123 |
"logging_steps": 10,
|
|
|
|
| 5137 |
"attributes": {}
|
| 5138 |
}
|
| 5139 |
},
|
| 5140 |
+
"total_flos": 7.127140008830661e+17,
|
| 5141 |
"train_batch_size": 2,
|
| 5142 |
"trial_name": null,
|
| 5143 |
"trial_params": null
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-4000 → checkpoint-7000}/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:aea6e1f55d32f940de995218d52f4fc064c98b8a4b1c3fe798b5213ef18bb2eb
|
| 3 |
+
size 166791
|
qwen3.5-2b-50pct-masked-weighted/test_eval_entity/metrics.json
CHANGED
|
@@ -1,14 +1,14 @@
|
|
| 1 |
{
|
| 2 |
-
"wer": 0.
|
| 3 |
-
"wer_zwnj_normalized": 0.
|
| 4 |
"exact_match": 0.0049504950495049506,
|
| 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": 0.005940594059405941,
|
| 14 |
"n_examples": 1010
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.3906661969876496,
|
| 3 |
+
"wer_zwnj_normalized": 0.3866045409421334,
|
| 4 |
"exact_match": 0.0049504950495049506,
|
| 5 |
+
"fix_rate": 0.2449823089206876,
|
| 6 |
+
"preservation_rate": 0.7061476596429146,
|
| 7 |
+
"targeted_score": 0.4755649842818011,
|
| 8 |
+
"fix_rate_lenient": 0.23776683087027914,
|
| 9 |
+
"preservation_rate_lenient": 0.7080494720500878,
|
| 10 |
+
"targeted_score_lenient": 0.4729081514601835,
|
| 11 |
+
"fix_rate_recoverable": 0.2412203027366592,
|
| 12 |
"unrecoverable_targets": 523,
|
| 13 |
"hallucination_rate": 0.005940594059405941,
|
| 14 |
"n_examples": 1010
|
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": 0.2560567766970463,
|
| 3 |
+
"wer_zwnj_normalized": 0.2507722549845549,
|
| 4 |
+
"exact_match": 0.0111731843575419,
|
| 5 |
+
"fix_rate": 0.3396665374176037,
|
| 6 |
+
"preservation_rate": 0.8318878076029753,
|
| 7 |
+
"targeted_score": 0.5857771725102895,
|
| 8 |
+
"fix_rate_lenient": 0.3351273028358518,
|
| 9 |
+
"preservation_rate_lenient": 0.8343051363942688,
|
| 10 |
+
"targeted_score_lenient": 0.5847162196150604,
|
| 11 |
+
"fix_rate_recoverable": 0.3396385035729298,
|
| 12 |
"unrecoverable_targets": 73,
|
| 13 |
+
"hallucination_rate": 0.008379888268156424,
|
| 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
|
|
|