checkpoint-3500
Browse files- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/README.md +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/adapter_config.json +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/adapter_model.safetensors +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/optimizer.pt +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/rng_state.pth +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/scheduler.pt +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/trainer_state.json +1164 -3
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/training_args.bin +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/tb/events.out.tfevents.1790605476.02b48645612f.4580.0 +2 -2
- qwen3.5-2b-100pct-masked-weighted-v2/test_eval_entity/metrics.json +12 -12
- qwen3.5-2b-100pct-masked-weighted-v2/test_eval_entity/predictions.jsonl +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/test_eval_typo/metrics.json +12 -12
- qwen3.5-2b-100pct-masked-weighted-v2/test_eval_typo/predictions.jsonl +0 -0
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/README.md
RENAMED
|
File without changes
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/adapter_config.json
RENAMED
|
File without changes
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/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 sha256:07f0bae6ceafa7326c80661d2a940ce36a3e7a0ae08971927456aa2c13dc23ae
|
| 3 |
size 134604152
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/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:b0cf8ee9f5204c6f6f56ac2d40ba79e3e3839494900204792175236114841aa5
|
| 3 |
size 269426431
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/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:e4f3d2751ddc8f607533e78b5b87d62031e61dec4e519e9b73091a08a04bb22a
|
| 3 |
size 14645
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/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:b4b33eea4bafd2e6601bcd171c7cfa6e732f7613557d6c95d65c8788701f9cf7
|
| 3 |
size 1465
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/trainer_state.json
RENAMED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 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,
|
|
@@ -1519,6 +1519,1167 @@
|
|
| 1519 |
"learning_rate": 6.751038911975822e-05,
|
| 1520 |
"loss": 0.7827093124389648,
|
| 1521 |
"step": 2000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1522 |
}
|
| 1523 |
],
|
| 1524 |
"logging_steps": 10,
|
|
@@ -1538,7 +2699,7 @@
|
|
| 1538 |
"attributes": {}
|
| 1539 |
}
|
| 1540 |
},
|
| 1541 |
-
"total_flos":
|
| 1542 |
"train_batch_size": 2,
|
| 1543 |
"trial_name": null,
|
| 1544 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.6280632100759284,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 3500,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1519 |
"learning_rate": 6.751038911975822e-05,
|
| 1520 |
"loss": 0.7827093124389648,
|
| 1521 |
"step": 2000
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.3588932629005305,
|
| 1525 |
+
"step": 2000,
|
| 1526 |
+
"test_entity_exact_match": 0.0029702970297029703,
|
| 1527 |
+
"test_entity_fix_rate": 0.24817609669035914,
|
| 1528 |
+
"test_entity_fix_rate_lenient": 0.23770114942528736,
|
| 1529 |
+
"test_entity_fix_rate_recoverable": 0.2408352513426065,
|
| 1530 |
+
"test_entity_hallucination_rate": 0.0019801980198019802,
|
| 1531 |
+
"test_entity_premature_stop_rate": 0.08415841584158416,
|
| 1532 |
+
"test_entity_preservation_rate": 0.7945375583078655,
|
| 1533 |
+
"test_entity_preservation_rate_lenient": 0.7966398521996844,
|
| 1534 |
+
"test_entity_targeted_score": 0.5213568274991124,
|
| 1535 |
+
"test_entity_targeted_score_lenient": 0.5171705008124858,
|
| 1536 |
+
"test_entity_wer": 0.315074561543963,
|
| 1537 |
+
"test_entity_wer_zwnj_normalized": 0.31014293186451897
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.3588932629005305,
|
| 1541 |
+
"step": 2000,
|
| 1542 |
+
"test_typo_exact_match": 0.008379888268156424,
|
| 1543 |
+
"test_typo_fix_rate": 0.34800310197751066,
|
| 1544 |
+
"test_typo_fix_rate_lenient": 0.3351273028358518,
|
| 1545 |
+
"test_typo_fix_rate_recoverable": 0.3394919168591224,
|
| 1546 |
+
"test_typo_hallucination_rate": 0.002793296089385475,
|
| 1547 |
+
"test_typo_premature_stop_rate": 0.0223463687150838,
|
| 1548 |
+
"test_typo_preservation_rate": 0.9328198041925146,
|
| 1549 |
+
"test_typo_preservation_rate_lenient": 0.9360954027366756,
|
| 1550 |
+
"test_typo_targeted_score": 0.6404114530850127,
|
| 1551 |
+
"test_typo_targeted_score_lenient": 0.6356113527862637,
|
| 1552 |
+
"test_typo_wer": 0.15692731867963544,
|
| 1553 |
+
"test_typo_wer_zwnj_normalized": 0.14972811518725587
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 0.3588932629005305,
|
| 1557 |
+
"step": 2000,
|
| 1558 |
+
"test_best_wer": 0.1826087192841977
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.3606877292150331,
|
| 1562 |
+
"grad_norm": 0.5443075895309448,
|
| 1563 |
+
"learning_rate": 6.732149603324519e-05,
|
| 1564 |
+
"loss": 0.6721854209899902,
|
| 1565 |
+
"step": 2010
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 0.3624821955295358,
|
| 1569 |
+
"grad_norm": 0.68841552734375,
|
| 1570 |
+
"learning_rate": 6.713260294673215e-05,
|
| 1571 |
+
"loss": 0.6306365489959717,
|
| 1572 |
+
"step": 2020
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 0.36427666184403845,
|
| 1576 |
+
"grad_norm": 0.7482131719589233,
|
| 1577 |
+
"learning_rate": 6.694370986021912e-05,
|
| 1578 |
+
"loss": 0.6596165657043457,
|
| 1579 |
+
"step": 2030
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 0.3660711281585411,
|
| 1583 |
+
"grad_norm": 1.0435492992401123,
|
| 1584 |
+
"learning_rate": 6.675481677370609e-05,
|
| 1585 |
+
"loss": 0.7504337787628174,
|
| 1586 |
+
"step": 2040
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.36786559447304373,
|
| 1590 |
+
"grad_norm": 2.6200437545776367,
|
| 1591 |
+
"learning_rate": 6.656592368719305e-05,
|
| 1592 |
+
"loss": 0.8442089080810546,
|
| 1593 |
+
"step": 2050
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.36966006078754643,
|
| 1597 |
+
"grad_norm": 0.5256720781326294,
|
| 1598 |
+
"learning_rate": 6.637703060068002e-05,
|
| 1599 |
+
"loss": 0.6658223152160645,
|
| 1600 |
+
"step": 2060
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 0.37145452710204907,
|
| 1604 |
+
"grad_norm": 0.6159254312515259,
|
| 1605 |
+
"learning_rate": 6.618813751416698e-05,
|
| 1606 |
+
"loss": 0.6340956211090087,
|
| 1607 |
+
"step": 2070
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 0.3732489934165517,
|
| 1611 |
+
"grad_norm": 0.7366557121276855,
|
| 1612 |
+
"learning_rate": 6.599924442765396e-05,
|
| 1613 |
+
"loss": 0.6326395511627197,
|
| 1614 |
+
"step": 2080
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.37504345973105435,
|
| 1618 |
+
"grad_norm": 1.0110527276992798,
|
| 1619 |
+
"learning_rate": 6.581035134114092e-05,
|
| 1620 |
+
"loss": 0.786867332458496,
|
| 1621 |
+
"step": 2090
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 0.376837926045557,
|
| 1625 |
+
"grad_norm": 2.832717180252075,
|
| 1626 |
+
"learning_rate": 6.562145825462789e-05,
|
| 1627 |
+
"loss": 0.8868856430053711,
|
| 1628 |
+
"step": 2100
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.3786323923600597,
|
| 1632 |
+
"grad_norm": 0.549351155757904,
|
| 1633 |
+
"learning_rate": 6.543256516811484e-05,
|
| 1634 |
+
"loss": 0.6318321228027344,
|
| 1635 |
+
"step": 2110
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.38042685867456233,
|
| 1639 |
+
"grad_norm": 0.6155033111572266,
|
| 1640 |
+
"learning_rate": 6.524367208160181e-05,
|
| 1641 |
+
"loss": 0.6392024993896485,
|
| 1642 |
+
"step": 2120
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.38222132498906497,
|
| 1646 |
+
"grad_norm": 0.6692315936088562,
|
| 1647 |
+
"learning_rate": 6.505477899508879e-05,
|
| 1648 |
+
"loss": 0.6711149215698242,
|
| 1649 |
+
"step": 2130
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 0.3840157913035676,
|
| 1653 |
+
"grad_norm": 0.9750716686248779,
|
| 1654 |
+
"learning_rate": 6.486588590857576e-05,
|
| 1655 |
+
"loss": 0.7159591674804687,
|
| 1656 |
+
"step": 2140
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 0.38581025761807025,
|
| 1660 |
+
"grad_norm": 2.3762757778167725,
|
| 1661 |
+
"learning_rate": 6.467699282206272e-05,
|
| 1662 |
+
"loss": 0.7706591129302979,
|
| 1663 |
+
"step": 2150
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 0.38760472393257295,
|
| 1667 |
+
"grad_norm": 0.5331564545631409,
|
| 1668 |
+
"learning_rate": 6.448809973554968e-05,
|
| 1669 |
+
"loss": 0.6141149997711182,
|
| 1670 |
+
"step": 2160
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.3893991902470756,
|
| 1674 |
+
"grad_norm": 0.6143497228622437,
|
| 1675 |
+
"learning_rate": 6.429920664903664e-05,
|
| 1676 |
+
"loss": 0.6400461196899414,
|
| 1677 |
+
"step": 2170
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.39119365656157823,
|
| 1681 |
+
"grad_norm": 0.7673105597496033,
|
| 1682 |
+
"learning_rate": 6.411031356252361e-05,
|
| 1683 |
+
"loss": 0.6552522659301758,
|
| 1684 |
+
"step": 2180
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.39298812287608087,
|
| 1688 |
+
"grad_norm": 0.986526608467102,
|
| 1689 |
+
"learning_rate": 6.392142047601059e-05,
|
| 1690 |
+
"loss": 0.6737249851226806,
|
| 1691 |
+
"step": 2190
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.3947825891905835,
|
| 1695 |
+
"grad_norm": 2.266063690185547,
|
| 1696 |
+
"learning_rate": 6.373252738949754e-05,
|
| 1697 |
+
"loss": 0.8194290161132812,
|
| 1698 |
+
"step": 2200
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 0.3965770555050862,
|
| 1702 |
+
"grad_norm": 0.5578099489212036,
|
| 1703 |
+
"learning_rate": 6.354363430298451e-05,
|
| 1704 |
+
"loss": 0.6300786972045899,
|
| 1705 |
+
"step": 2210
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 0.39837152181958885,
|
| 1709 |
+
"grad_norm": 0.6543112993240356,
|
| 1710 |
+
"learning_rate": 6.335474121647148e-05,
|
| 1711 |
+
"loss": 0.6619680881500244,
|
| 1712 |
+
"step": 2220
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.4001659881340915,
|
| 1716 |
+
"grad_norm": 0.8078407645225525,
|
| 1717 |
+
"learning_rate": 6.316584812995844e-05,
|
| 1718 |
+
"loss": 0.6784715175628662,
|
| 1719 |
+
"step": 2230
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.40196045444859413,
|
| 1723 |
+
"grad_norm": 0.9718537926673889,
|
| 1724 |
+
"learning_rate": 6.297695504344541e-05,
|
| 1725 |
+
"loss": 0.7365913391113281,
|
| 1726 |
+
"step": 2240
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 0.4037549207630968,
|
| 1730 |
+
"grad_norm": 2.394310712814331,
|
| 1731 |
+
"learning_rate": 6.278806195693238e-05,
|
| 1732 |
+
"loss": 0.8184401512145996,
|
| 1733 |
+
"step": 2250
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 0.40554938707759947,
|
| 1737 |
+
"grad_norm": 0.518998920917511,
|
| 1738 |
+
"learning_rate": 6.259916887041935e-05,
|
| 1739 |
+
"loss": 0.6356068611145019,
|
| 1740 |
+
"step": 2260
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 0.4073438533921021,
|
| 1744 |
+
"grad_norm": 0.6347622275352478,
|
| 1745 |
+
"learning_rate": 6.241027578390631e-05,
|
| 1746 |
+
"loss": 0.642417049407959,
|
| 1747 |
+
"step": 2270
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 0.40913831970660475,
|
| 1751 |
+
"grad_norm": 0.708690881729126,
|
| 1752 |
+
"learning_rate": 6.222138269739328e-05,
|
| 1753 |
+
"loss": 0.6165683269500732,
|
| 1754 |
+
"step": 2280
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.4109327860211074,
|
| 1758 |
+
"grad_norm": 0.9712520241737366,
|
| 1759 |
+
"learning_rate": 6.203248961088025e-05,
|
| 1760 |
+
"loss": 0.7352517127990723,
|
| 1761 |
+
"step": 2290
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.4127272523356101,
|
| 1765 |
+
"grad_norm": 2.8482367992401123,
|
| 1766 |
+
"learning_rate": 6.184359652436721e-05,
|
| 1767 |
+
"loss": 0.8003445625305176,
|
| 1768 |
+
"step": 2300
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.4145217186501127,
|
| 1772 |
+
"grad_norm": 0.5100517868995667,
|
| 1773 |
+
"learning_rate": 6.165470343785418e-05,
|
| 1774 |
+
"loss": 0.6422919273376465,
|
| 1775 |
+
"step": 2310
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 0.41631618496461537,
|
| 1779 |
+
"grad_norm": 0.6365388631820679,
|
| 1780 |
+
"learning_rate": 6.146581035134115e-05,
|
| 1781 |
+
"loss": 0.6152407169342041,
|
| 1782 |
+
"step": 2320
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 0.418110651279118,
|
| 1786 |
+
"grad_norm": 0.7218090891838074,
|
| 1787 |
+
"learning_rate": 6.127691726482811e-05,
|
| 1788 |
+
"loss": 0.6350100994110107,
|
| 1789 |
+
"step": 2330
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 0.41990511759362065,
|
| 1793 |
+
"grad_norm": 1.0017449855804443,
|
| 1794 |
+
"learning_rate": 6.108802417831508e-05,
|
| 1795 |
+
"loss": 0.7071941375732422,
|
| 1796 |
+
"step": 2340
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.42169958390812334,
|
| 1800 |
+
"grad_norm": 2.7050940990448,
|
| 1801 |
+
"learning_rate": 6.089913109180204e-05,
|
| 1802 |
+
"loss": 0.8486111640930176,
|
| 1803 |
+
"step": 2350
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 0.423494050222626,
|
| 1807 |
+
"grad_norm": 0.4976552724838257,
|
| 1808 |
+
"learning_rate": 6.0710238005289014e-05,
|
| 1809 |
+
"loss": 0.6660837173461914,
|
| 1810 |
+
"step": 2360
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.4252885165371286,
|
| 1814 |
+
"grad_norm": 0.6267609000205994,
|
| 1815 |
+
"learning_rate": 6.052134491877598e-05,
|
| 1816 |
+
"loss": 0.6445472717285157,
|
| 1817 |
+
"step": 2370
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 0.42708298285163127,
|
| 1821 |
+
"grad_norm": 0.781930148601532,
|
| 1822 |
+
"learning_rate": 6.033245183226294e-05,
|
| 1823 |
+
"loss": 0.6631790161132812,
|
| 1824 |
+
"step": 2380
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.42887744916613396,
|
| 1828 |
+
"grad_norm": 0.9512560367584229,
|
| 1829 |
+
"learning_rate": 6.014355874574991e-05,
|
| 1830 |
+
"loss": 0.7210745334625244,
|
| 1831 |
+
"step": 2390
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.4306719154806366,
|
| 1835 |
+
"grad_norm": 2.8436152935028076,
|
| 1836 |
+
"learning_rate": 5.9954665659236874e-05,
|
| 1837 |
+
"loss": 0.8688377380371094,
|
| 1838 |
+
"step": 2400
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.43246638179513924,
|
| 1842 |
+
"grad_norm": 0.5824747085571289,
|
| 1843 |
+
"learning_rate": 5.976577257272384e-05,
|
| 1844 |
+
"loss": 0.6362675189971924,
|
| 1845 |
+
"step": 2410
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 0.4342608481096419,
|
| 1849 |
+
"grad_norm": 0.6418686509132385,
|
| 1850 |
+
"learning_rate": 5.9576879486210815e-05,
|
| 1851 |
+
"loss": 0.6117087364196777,
|
| 1852 |
+
"step": 2420
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 0.4360553144241445,
|
| 1856 |
+
"grad_norm": 0.8100792169570923,
|
| 1857 |
+
"learning_rate": 5.938798639969777e-05,
|
| 1858 |
+
"loss": 0.6724053382873535,
|
| 1859 |
+
"step": 2430
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.4378497807386472,
|
| 1863 |
+
"grad_norm": 1.0161293745040894,
|
| 1864 |
+
"learning_rate": 5.919909331318474e-05,
|
| 1865 |
+
"loss": 0.725592851638794,
|
| 1866 |
+
"step": 2440
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.43964424705314986,
|
| 1870 |
+
"grad_norm": 3.063896894454956,
|
| 1871 |
+
"learning_rate": 5.901020022667171e-05,
|
| 1872 |
+
"loss": 0.7809360027313232,
|
| 1873 |
+
"step": 2450
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.4414387133676525,
|
| 1877 |
+
"grad_norm": 0.5110781192779541,
|
| 1878 |
+
"learning_rate": 5.8821307140158675e-05,
|
| 1879 |
+
"loss": 0.6253266334533691,
|
| 1880 |
+
"step": 2460
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.44323317968215514,
|
| 1884 |
+
"grad_norm": 0.6076716780662537,
|
| 1885 |
+
"learning_rate": 5.8632414053645635e-05,
|
| 1886 |
+
"loss": 0.6204933643341064,
|
| 1887 |
+
"step": 2470
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 0.4450276459966578,
|
| 1891 |
+
"grad_norm": 0.7593351602554321,
|
| 1892 |
+
"learning_rate": 5.84435209671326e-05,
|
| 1893 |
+
"loss": 0.645332145690918,
|
| 1894 |
+
"step": 2480
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 0.4468221123111605,
|
| 1898 |
+
"grad_norm": 1.0506312847137451,
|
| 1899 |
+
"learning_rate": 5.825462788061957e-05,
|
| 1900 |
+
"loss": 0.700313949584961,
|
| 1901 |
+
"step": 2490
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 0.4486165786256631,
|
| 1905 |
+
"grad_norm": 2.1724884510040283,
|
| 1906 |
+
"learning_rate": 5.806573479410654e-05,
|
| 1907 |
+
"loss": 0.7738859176635742,
|
| 1908 |
+
"step": 2500
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 0.4486165786256631,
|
| 1912 |
+
"step": 2500,
|
| 1913 |
+
"test_entity_exact_match": 0.0039603960396039604,
|
| 1914 |
+
"test_entity_fix_rate": 0.2834330087359489,
|
| 1915 |
+
"test_entity_fix_rate_lenient": 0.2722824302134647,
|
| 1916 |
+
"test_entity_fix_rate_recoverable": 0.27582641182160844,
|
| 1917 |
+
"test_entity_hallucination_rate": 0.0,
|
| 1918 |
+
"test_entity_premature_stop_rate": 0.013861386138613862,
|
| 1919 |
+
"test_entity_preservation_rate": 0.9140485764838346,
|
| 1920 |
+
"test_entity_preservation_rate_lenient": 0.9170034512400087,
|
| 1921 |
+
"test_entity_targeted_score": 0.5987407926098918,
|
| 1922 |
+
"test_entity_targeted_score_lenient": 0.5946429407267367,
|
| 1923 |
+
"test_entity_wer": 0.2125244179716271,
|
| 1924 |
+
"test_entity_wer_zwnj_normalized": 0.20712310927509314
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 0.4486165786256631,
|
| 1928 |
+
"step": 2500,
|
| 1929 |
+
"test_typo_exact_match": 0.0111731843575419,
|
| 1930 |
+
"test_typo_fix_rate": 0.35440093059325317,
|
| 1931 |
+
"test_typo_fix_rate_lenient": 0.3454771268888429,
|
| 1932 |
+
"test_typo_fix_rate_recoverable": 0.34998950241444465,
|
| 1933 |
+
"test_typo_hallucination_rate": 0.002793296089385475,
|
| 1934 |
+
"test_typo_premature_stop_rate": 0.0,
|
| 1935 |
+
"test_typo_preservation_rate": 0.9698056625407933,
|
| 1936 |
+
"test_typo_preservation_rate_lenient": 0.9731899557561019,
|
| 1937 |
+
"test_typo_targeted_score": 0.6621032965670233,
|
| 1938 |
+
"test_typo_targeted_score_lenient": 0.6593335413224723,
|
| 1939 |
+
"test_typo_wer": 0.1239947920655587,
|
| 1940 |
+
"test_typo_wer_zwnj_normalized": 0.1172295831099538
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.4486165786256631,
|
| 1944 |
+
"step": 2500,
|
| 1945 |
+
"test_best_wer": 0.1682596050185929
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 0.45041104494016576,
|
| 1949 |
+
"grad_norm": 0.534072756767273,
|
| 1950 |
+
"learning_rate": 5.787684170759351e-05,
|
| 1951 |
+
"loss": 0.6257661342620849,
|
| 1952 |
+
"step": 2510
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.4522055112546684,
|
| 1956 |
+
"grad_norm": 0.6417807936668396,
|
| 1957 |
+
"learning_rate": 5.768794862108047e-05,
|
| 1958 |
+
"loss": 0.6577399253845215,
|
| 1959 |
+
"step": 2520
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.45399997756917104,
|
| 1963 |
+
"grad_norm": 0.740066647529602,
|
| 1964 |
+
"learning_rate": 5.7499055534567436e-05,
|
| 1965 |
+
"loss": 0.6428184509277344,
|
| 1966 |
+
"step": 2530
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 0.45579444388367374,
|
| 1970 |
+
"grad_norm": 1.027453064918518,
|
| 1971 |
+
"learning_rate": 5.73101624480544e-05,
|
| 1972 |
+
"loss": 0.7034515857696533,
|
| 1973 |
+
"step": 2540
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 0.4575889101981764,
|
| 1977 |
+
"grad_norm": 2.817686080932617,
|
| 1978 |
+
"learning_rate": 5.712126936154138e-05,
|
| 1979 |
+
"loss": 0.791818380355835,
|
| 1980 |
+
"step": 2550
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 0.459383376512679,
|
| 1984 |
+
"grad_norm": 0.613205075263977,
|
| 1985 |
+
"learning_rate": 5.693237627502833e-05,
|
| 1986 |
+
"loss": 0.6438217639923096,
|
| 1987 |
+
"step": 2560
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 0.46117784282718166,
|
| 1991 |
+
"grad_norm": 0.5848596096038818,
|
| 1992 |
+
"learning_rate": 5.6743483188515304e-05,
|
| 1993 |
+
"loss": 0.5878115177154541,
|
| 1994 |
+
"step": 2570
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.46297230914168436,
|
| 1998 |
+
"grad_norm": 0.733321487903595,
|
| 1999 |
+
"learning_rate": 5.655459010200227e-05,
|
| 2000 |
+
"loss": 0.6228268146514893,
|
| 2001 |
+
"step": 2580
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.464766775456187,
|
| 2005 |
+
"grad_norm": 0.9225145578384399,
|
| 2006 |
+
"learning_rate": 5.636569701548924e-05,
|
| 2007 |
+
"loss": 0.7020124912261962,
|
| 2008 |
+
"step": 2590
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.46656124177068964,
|
| 2012 |
+
"grad_norm": 2.770216941833496,
|
| 2013 |
+
"learning_rate": 5.6176803928976204e-05,
|
| 2014 |
+
"loss": 0.8512001991271972,
|
| 2015 |
+
"step": 2600
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 0.4683557080851923,
|
| 2019 |
+
"grad_norm": 0.5025762319564819,
|
| 2020 |
+
"learning_rate": 5.5987910842463164e-05,
|
| 2021 |
+
"loss": 0.5970348834991455,
|
| 2022 |
+
"step": 2610
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 0.4701501743996949,
|
| 2026 |
+
"grad_norm": 0.6208140254020691,
|
| 2027 |
+
"learning_rate": 5.579901775595013e-05,
|
| 2028 |
+
"loss": 0.6049718856811523,
|
| 2029 |
+
"step": 2620
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 0.4719446407141976,
|
| 2033 |
+
"grad_norm": 0.749312698841095,
|
| 2034 |
+
"learning_rate": 5.5610124669437105e-05,
|
| 2035 |
+
"loss": 0.6626112937927247,
|
| 2036 |
+
"step": 2630
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.47373910702870026,
|
| 2040 |
+
"grad_norm": 1.0702252388000488,
|
| 2041 |
+
"learning_rate": 5.542123158292407e-05,
|
| 2042 |
+
"loss": 0.7458027839660645,
|
| 2043 |
+
"step": 2640
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 0.4755335733432029,
|
| 2047 |
+
"grad_norm": 2.6121034622192383,
|
| 2048 |
+
"learning_rate": 5.523233849641103e-05,
|
| 2049 |
+
"loss": 0.7557179450988769,
|
| 2050 |
+
"step": 2650
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 0.47732803965770554,
|
| 2054 |
+
"grad_norm": 0.5720552802085876,
|
| 2055 |
+
"learning_rate": 5.5043445409898e-05,
|
| 2056 |
+
"loss": 0.6178184986114502,
|
| 2057 |
+
"step": 2660
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 0.4791225059722082,
|
| 2061 |
+
"grad_norm": 0.6336584687232971,
|
| 2062 |
+
"learning_rate": 5.4854552323384965e-05,
|
| 2063 |
+
"loss": 0.6276357650756836,
|
| 2064 |
+
"step": 2670
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 0.4809169722867109,
|
| 2068 |
+
"grad_norm": 0.7830484509468079,
|
| 2069 |
+
"learning_rate": 5.466565923687194e-05,
|
| 2070 |
+
"loss": 0.6212595462799072,
|
| 2071 |
+
"step": 2680
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 0.4827114386012135,
|
| 2075 |
+
"grad_norm": 1.01641047000885,
|
| 2076 |
+
"learning_rate": 5.4476766150358906e-05,
|
| 2077 |
+
"loss": 0.6875291347503663,
|
| 2078 |
+
"step": 2690
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.48450590491571616,
|
| 2082 |
+
"grad_norm": 3.129596471786499,
|
| 2083 |
+
"learning_rate": 5.428787306384586e-05,
|
| 2084 |
+
"loss": 0.8276726722717285,
|
| 2085 |
+
"step": 2700
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 0.4863003712302188,
|
| 2089 |
+
"grad_norm": 0.5203351974487305,
|
| 2090 |
+
"learning_rate": 5.409897997733283e-05,
|
| 2091 |
+
"loss": 0.6424898624420166,
|
| 2092 |
+
"step": 2710
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 0.48809483754472144,
|
| 2096 |
+
"grad_norm": 0.6404901742935181,
|
| 2097 |
+
"learning_rate": 5.39100868908198e-05,
|
| 2098 |
+
"loss": 0.6241347312927246,
|
| 2099 |
+
"step": 2720
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 0.48988930385922413,
|
| 2103 |
+
"grad_norm": 0.7382896542549133,
|
| 2104 |
+
"learning_rate": 5.3721193804306766e-05,
|
| 2105 |
+
"loss": 0.6695589542388916,
|
| 2106 |
+
"step": 2730
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 0.4916837701737268,
|
| 2110 |
+
"grad_norm": 1.0695264339447021,
|
| 2111 |
+
"learning_rate": 5.3532300717793726e-05,
|
| 2112 |
+
"loss": 0.7361619472503662,
|
| 2113 |
+
"step": 2740
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 0.4934782364882294,
|
| 2117 |
+
"grad_norm": 3.029681921005249,
|
| 2118 |
+
"learning_rate": 5.334340763128069e-05,
|
| 2119 |
+
"loss": 0.7602876186370849,
|
| 2120 |
+
"step": 2750
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 0.49527270280273206,
|
| 2124 |
+
"grad_norm": 0.6055657267570496,
|
| 2125 |
+
"learning_rate": 5.315451454476767e-05,
|
| 2126 |
+
"loss": 0.64903883934021,
|
| 2127 |
+
"step": 2760
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 0.49706716911723475,
|
| 2131 |
+
"grad_norm": 0.6373789310455322,
|
| 2132 |
+
"learning_rate": 5.2965621458254634e-05,
|
| 2133 |
+
"loss": 0.6150396347045899,
|
| 2134 |
+
"step": 2770
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.4988616354317374,
|
| 2138 |
+
"grad_norm": 0.7586564421653748,
|
| 2139 |
+
"learning_rate": 5.27767283717416e-05,
|
| 2140 |
+
"loss": 0.6315049171447754,
|
| 2141 |
+
"step": 2780
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.50065610174624,
|
| 2145 |
+
"grad_norm": 0.9273103475570679,
|
| 2146 |
+
"learning_rate": 5.258783528522856e-05,
|
| 2147 |
+
"loss": 0.7004617691040039,
|
| 2148 |
+
"step": 2790
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 0.5024505680607427,
|
| 2152 |
+
"grad_norm": 3.1377556324005127,
|
| 2153 |
+
"learning_rate": 5.239894219871553e-05,
|
| 2154 |
+
"loss": 0.8075312614440918,
|
| 2155 |
+
"step": 2800
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 0.5042450343752454,
|
| 2159 |
+
"grad_norm": 0.549586296081543,
|
| 2160 |
+
"learning_rate": 5.2210049112202494e-05,
|
| 2161 |
+
"loss": 0.5876753807067872,
|
| 2162 |
+
"step": 2810
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.506039500689748,
|
| 2166 |
+
"grad_norm": 0.6282278895378113,
|
| 2167 |
+
"learning_rate": 5.202115602568947e-05,
|
| 2168 |
+
"loss": 0.6268801212310791,
|
| 2169 |
+
"step": 2820
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.5078339670042507,
|
| 2173 |
+
"grad_norm": 0.8212598562240601,
|
| 2174 |
+
"learning_rate": 5.183226293917642e-05,
|
| 2175 |
+
"loss": 0.642136526107788,
|
| 2176 |
+
"step": 2830
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 0.5096284333187533,
|
| 2180 |
+
"grad_norm": 0.969655454158783,
|
| 2181 |
+
"learning_rate": 5.1643369852663395e-05,
|
| 2182 |
+
"loss": 0.7123368740081787,
|
| 2183 |
+
"step": 2840
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 0.5114228996332559,
|
| 2187 |
+
"grad_norm": 2.6947407722473145,
|
| 2188 |
+
"learning_rate": 5.145447676615036e-05,
|
| 2189 |
+
"loss": 0.7797284603118897,
|
| 2190 |
+
"step": 2850
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 0.5132173659477586,
|
| 2194 |
+
"grad_norm": 0.5469926595687866,
|
| 2195 |
+
"learning_rate": 5.126558367963733e-05,
|
| 2196 |
+
"loss": 0.6178148746490478,
|
| 2197 |
+
"step": 2860
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 0.5150118322622612,
|
| 2201 |
+
"grad_norm": 0.6259074807167053,
|
| 2202 |
+
"learning_rate": 5.10766905931243e-05,
|
| 2203 |
+
"loss": 0.6211585521697998,
|
| 2204 |
+
"step": 2870
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.5168062985767639,
|
| 2208 |
+
"grad_norm": 0.7744222283363342,
|
| 2209 |
+
"learning_rate": 5.0887797506611256e-05,
|
| 2210 |
+
"loss": 0.6576568603515625,
|
| 2211 |
+
"step": 2880
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 0.5186007648912666,
|
| 2215 |
+
"grad_norm": 1.0394456386566162,
|
| 2216 |
+
"learning_rate": 5.069890442009822e-05,
|
| 2217 |
+
"loss": 0.683222484588623,
|
| 2218 |
+
"step": 2890
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.5203952312057692,
|
| 2222 |
+
"grad_norm": 2.6866447925567627,
|
| 2223 |
+
"learning_rate": 5.0510011333585196e-05,
|
| 2224 |
+
"loss": 0.7875309944152832,
|
| 2225 |
+
"step": 2900
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 0.5221896975202719,
|
| 2229 |
+
"grad_norm": 0.5467143058776855,
|
| 2230 |
+
"learning_rate": 5.032111824707216e-05,
|
| 2231 |
+
"loss": 0.6157989978790284,
|
| 2232 |
+
"step": 2910
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 0.5239841638347745,
|
| 2236 |
+
"grad_norm": 0.5998079776763916,
|
| 2237 |
+
"learning_rate": 5.013222516055912e-05,
|
| 2238 |
+
"loss": 0.604616355895996,
|
| 2239 |
+
"step": 2920
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 0.5257786301492772,
|
| 2243 |
+
"grad_norm": 0.8082038164138794,
|
| 2244 |
+
"learning_rate": 4.994333207404609e-05,
|
| 2245 |
+
"loss": 0.6244801998138427,
|
| 2246 |
+
"step": 2930
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.5275730964637798,
|
| 2250 |
+
"grad_norm": 1.0210673809051514,
|
| 2251 |
+
"learning_rate": 4.9754438987533057e-05,
|
| 2252 |
+
"loss": 0.7031500816345215,
|
| 2253 |
+
"step": 2940
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 0.5293675627782825,
|
| 2257 |
+
"grad_norm": 2.866973876953125,
|
| 2258 |
+
"learning_rate": 4.956554590102003e-05,
|
| 2259 |
+
"loss": 0.777482271194458,
|
| 2260 |
+
"step": 2950
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.5311620290927851,
|
| 2264 |
+
"grad_norm": 0.5310227870941162,
|
| 2265 |
+
"learning_rate": 4.937665281450699e-05,
|
| 2266 |
+
"loss": 0.6212926864624023,
|
| 2267 |
+
"step": 2960
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 0.5329564954072877,
|
| 2271 |
+
"grad_norm": 0.6013730764389038,
|
| 2272 |
+
"learning_rate": 4.918775972799396e-05,
|
| 2273 |
+
"loss": 0.6015076637268066,
|
| 2274 |
+
"step": 2970
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.5347509617217904,
|
| 2278 |
+
"grad_norm": 0.7386709451675415,
|
| 2279 |
+
"learning_rate": 4.8998866641480924e-05,
|
| 2280 |
+
"loss": 0.6311803817749023,
|
| 2281 |
+
"step": 2980
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 0.5365454280362931,
|
| 2285 |
+
"grad_norm": 1.0695222616195679,
|
| 2286 |
+
"learning_rate": 4.880997355496789e-05,
|
| 2287 |
+
"loss": 0.7098263740539551,
|
| 2288 |
+
"step": 2990
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.5383398943507958,
|
| 2292 |
+
"grad_norm": 2.4510560035705566,
|
| 2293 |
+
"learning_rate": 4.862108046845486e-05,
|
| 2294 |
+
"loss": 0.7919962406158447,
|
| 2295 |
+
"step": 3000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 0.5383398943507958,
|
| 2299 |
+
"step": 3000,
|
| 2300 |
+
"test_entity_exact_match": 0.0049504950495049506,
|
| 2301 |
+
"test_entity_fix_rate": 0.23283339073801546,
|
| 2302 |
+
"test_entity_fix_rate_lenient": 0.22308702791461413,
|
| 2303 |
+
"test_entity_fix_rate_recoverable": 0.22609159745154941,
|
| 2304 |
+
"test_entity_hallucination_rate": 0.0039603960396039604,
|
| 2305 |
+
"test_entity_premature_stop_rate": 0.14455445544554454,
|
| 2306 |
+
"test_entity_preservation_rate": 0.7337622647579218,
|
| 2307 |
+
"test_entity_preservation_rate_lenient": 0.7359865542768433,
|
| 2308 |
+
"test_entity_targeted_score": 0.48329782774796864,
|
| 2309 |
+
"test_entity_targeted_score_lenient": 0.4795367910957287,
|
| 2310 |
+
"test_entity_wer": 0.36886348352387355,
|
| 2311 |
+
"test_entity_wer_zwnj_normalized": 0.364348160245941
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 0.5383398943507958,
|
| 2315 |
+
"step": 3000,
|
| 2316 |
+
"test_typo_exact_match": 0.019553072625698324,
|
| 2317 |
+
"test_typo_fix_rate": 0.3286157425358666,
|
| 2318 |
+
"test_typo_fix_rate_lenient": 0.3181535913889464,
|
| 2319 |
+
"test_typo_fix_rate_recoverable": 0.32227587654839385,
|
| 2320 |
+
"test_typo_hallucination_rate": 0.0111731843575419,
|
| 2321 |
+
"test_typo_premature_stop_rate": 0.06983240223463687,
|
| 2322 |
+
"test_typo_preservation_rate": 0.8671684355981536,
|
| 2323 |
+
"test_typo_preservation_rate_lenient": 0.8702862668112162,
|
| 2324 |
+
"test_typo_targeted_score": 0.5978920890670101,
|
| 2325 |
+
"test_typo_targeted_score_lenient": 0.5942199291000814,
|
| 2326 |
+
"test_typo_wer": 0.2218733246534426,
|
| 2327 |
+
"test_typo_wer_zwnj_normalized": 0.21566975568660487
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.5383398943507958,
|
| 2331 |
+
"step": 3000,
|
| 2332 |
+
"test_best_wer": 0.1682596050185929
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.5401343606652984,
|
| 2336 |
+
"grad_norm": 0.5469890832901001,
|
| 2337 |
+
"learning_rate": 4.8432187381941824e-05,
|
| 2338 |
+
"loss": 0.6426947593688965,
|
| 2339 |
+
"step": 3010
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 0.541928826979801,
|
| 2343 |
+
"grad_norm": 0.6795536875724792,
|
| 2344 |
+
"learning_rate": 4.8243294295428785e-05,
|
| 2345 |
+
"loss": 0.6286990642547607,
|
| 2346 |
+
"step": 3020
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 0.5437232932943037,
|
| 2350 |
+
"grad_norm": 0.7367618680000305,
|
| 2351 |
+
"learning_rate": 4.805440120891576e-05,
|
| 2352 |
+
"loss": 0.6102496147155761,
|
| 2353 |
+
"step": 3030
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 0.5455177596088063,
|
| 2357 |
+
"grad_norm": 1.017337441444397,
|
| 2358 |
+
"learning_rate": 4.7865508122402725e-05,
|
| 2359 |
+
"loss": 0.6789227485656738,
|
| 2360 |
+
"step": 3040
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.547312225923309,
|
| 2364 |
+
"grad_norm": 2.496983289718628,
|
| 2365 |
+
"learning_rate": 4.7676615035889685e-05,
|
| 2366 |
+
"loss": 0.8142426490783692,
|
| 2367 |
+
"step": 3050
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 0.5491066922378116,
|
| 2371 |
+
"grad_norm": 0.6170143485069275,
|
| 2372 |
+
"learning_rate": 4.748772194937666e-05,
|
| 2373 |
+
"loss": 0.617520809173584,
|
| 2374 |
+
"step": 3060
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 0.5509011585523143,
|
| 2378 |
+
"grad_norm": 0.6180261373519897,
|
| 2379 |
+
"learning_rate": 4.729882886286362e-05,
|
| 2380 |
+
"loss": 0.5764644145965576,
|
| 2381 |
+
"step": 3070
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 0.552695624866817,
|
| 2385 |
+
"grad_norm": 0.7875624895095825,
|
| 2386 |
+
"learning_rate": 4.7109935776350586e-05,
|
| 2387 |
+
"loss": 0.6153441429138183,
|
| 2388 |
+
"step": 3080
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 0.5544900911813196,
|
| 2392 |
+
"grad_norm": 1.011209487915039,
|
| 2393 |
+
"learning_rate": 4.692104268983755e-05,
|
| 2394 |
+
"loss": 0.7133022308349609,
|
| 2395 |
+
"step": 3090
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 0.5562845574958223,
|
| 2399 |
+
"grad_norm": 3.032257318496704,
|
| 2400 |
+
"learning_rate": 4.673214960332452e-05,
|
| 2401 |
+
"loss": 0.805880355834961,
|
| 2402 |
+
"step": 3100
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 0.5580790238103249,
|
| 2406 |
+
"grad_norm": 0.6044086217880249,
|
| 2407 |
+
"learning_rate": 4.6543256516811486e-05,
|
| 2408 |
+
"loss": 0.6557058334350586,
|
| 2409 |
+
"step": 3110
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 0.5598734901248276,
|
| 2413 |
+
"grad_norm": 0.6103405952453613,
|
| 2414 |
+
"learning_rate": 4.635436343029845e-05,
|
| 2415 |
+
"loss": 0.5969423770904541,
|
| 2416 |
+
"step": 3120
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 0.5616679564393302,
|
| 2420 |
+
"grad_norm": 0.7534403204917908,
|
| 2421 |
+
"learning_rate": 4.616547034378542e-05,
|
| 2422 |
+
"loss": 0.6017137050628663,
|
| 2423 |
+
"step": 3130
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 0.5634624227538328,
|
| 2427 |
+
"grad_norm": 1.0295764207839966,
|
| 2428 |
+
"learning_rate": 4.597657725727239e-05,
|
| 2429 |
+
"loss": 0.6981183528900147,
|
| 2430 |
+
"step": 3140
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 0.5652568890683355,
|
| 2434 |
+
"grad_norm": 3.2020111083984375,
|
| 2435 |
+
"learning_rate": 4.5787684170759354e-05,
|
| 2436 |
+
"loss": 0.7978395462036133,
|
| 2437 |
+
"step": 3150
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 0.5670513553828381,
|
| 2441 |
+
"grad_norm": 0.568731427192688,
|
| 2442 |
+
"learning_rate": 4.559879108424632e-05,
|
| 2443 |
+
"loss": 0.5992100715637207,
|
| 2444 |
+
"step": 3160
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 0.5688458216973408,
|
| 2448 |
+
"grad_norm": 0.6443565487861633,
|
| 2449 |
+
"learning_rate": 4.540989799773329e-05,
|
| 2450 |
+
"loss": 0.646563196182251,
|
| 2451 |
+
"step": 3170
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 0.5706402880118435,
|
| 2455 |
+
"grad_norm": 0.7451872825622559,
|
| 2456 |
+
"learning_rate": 4.522100491122025e-05,
|
| 2457 |
+
"loss": 0.615162706375122,
|
| 2458 |
+
"step": 3180
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.5724347543263462,
|
| 2462 |
+
"grad_norm": 1.0265822410583496,
|
| 2463 |
+
"learning_rate": 4.503211182470722e-05,
|
| 2464 |
+
"loss": 0.6840673923492432,
|
| 2465 |
+
"step": 3190
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 0.5742292206408488,
|
| 2469 |
+
"grad_norm": 3.0063066482543945,
|
| 2470 |
+
"learning_rate": 4.484321873819418e-05,
|
| 2471 |
+
"loss": 0.7510017395019531,
|
| 2472 |
+
"step": 3200
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 0.5760236869553514,
|
| 2476 |
+
"grad_norm": 0.5863959193229675,
|
| 2477 |
+
"learning_rate": 4.465432565168115e-05,
|
| 2478 |
+
"loss": 0.6135642051696777,
|
| 2479 |
+
"step": 3210
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 0.5778181532698541,
|
| 2483 |
+
"grad_norm": 0.6326755881309509,
|
| 2484 |
+
"learning_rate": 4.446543256516812e-05,
|
| 2485 |
+
"loss": 0.5748859882354737,
|
| 2486 |
+
"step": 3220
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.5796126195843567,
|
| 2490 |
+
"grad_norm": 0.8056603074073792,
|
| 2491 |
+
"learning_rate": 4.427653947865508e-05,
|
| 2492 |
+
"loss": 0.6199123859405518,
|
| 2493 |
+
"step": 3230
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 0.5814070858988594,
|
| 2497 |
+
"grad_norm": 1.0059834718704224,
|
| 2498 |
+
"learning_rate": 4.408764639214205e-05,
|
| 2499 |
+
"loss": 0.7042075157165527,
|
| 2500 |
+
"step": 3240
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 0.583201552213362,
|
| 2504 |
+
"grad_norm": 2.5240535736083984,
|
| 2505 |
+
"learning_rate": 4.3898753305629015e-05,
|
| 2506 |
+
"loss": 0.7729485511779786,
|
| 2507 |
+
"step": 3250
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 0.5849960185278646,
|
| 2511 |
+
"grad_norm": 0.6064797639846802,
|
| 2512 |
+
"learning_rate": 4.370986021911598e-05,
|
| 2513 |
+
"loss": 0.5992433071136475,
|
| 2514 |
+
"step": 3260
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 0.5867904848423674,
|
| 2518 |
+
"grad_norm": 0.6692094802856445,
|
| 2519 |
+
"learning_rate": 4.352096713260295e-05,
|
| 2520 |
+
"loss": 0.6091554641723633,
|
| 2521 |
+
"step": 3270
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 0.58858495115687,
|
| 2525 |
+
"grad_norm": 0.8525553941726685,
|
| 2526 |
+
"learning_rate": 4.3332074046089916e-05,
|
| 2527 |
+
"loss": 0.5793965816497803,
|
| 2528 |
+
"step": 3280
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 0.5903794174713727,
|
| 2532 |
+
"grad_norm": 1.0216304063796997,
|
| 2533 |
+
"learning_rate": 4.314318095957688e-05,
|
| 2534 |
+
"loss": 0.6783274650573731,
|
| 2535 |
+
"step": 3290
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 0.5921738837858753,
|
| 2539 |
+
"grad_norm": 3.011688232421875,
|
| 2540 |
+
"learning_rate": 4.295428787306385e-05,
|
| 2541 |
+
"loss": 0.7689754486083984,
|
| 2542 |
+
"step": 3300
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 0.593968350100378,
|
| 2546 |
+
"grad_norm": 0.5574772953987122,
|
| 2547 |
+
"learning_rate": 4.2765394786550816e-05,
|
| 2548 |
+
"loss": 0.5883103847503662,
|
| 2549 |
+
"step": 3310
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 0.5957628164148806,
|
| 2553 |
+
"grad_norm": 0.6558288335800171,
|
| 2554 |
+
"learning_rate": 4.2576501700037776e-05,
|
| 2555 |
+
"loss": 0.6355991840362549,
|
| 2556 |
+
"step": 3320
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 0.5975572827293832,
|
| 2560 |
+
"grad_norm": 0.7502368688583374,
|
| 2561 |
+
"learning_rate": 4.238760861352475e-05,
|
| 2562 |
+
"loss": 0.6298327445983887,
|
| 2563 |
+
"step": 3330
|
| 2564 |
+
},
|
| 2565 |
+
{
|
| 2566 |
+
"epoch": 0.5993517490438859,
|
| 2567 |
+
"grad_norm": 1.0399233102798462,
|
| 2568 |
+
"learning_rate": 4.219871552701171e-05,
|
| 2569 |
+
"loss": 0.6809325218200684,
|
| 2570 |
+
"step": 3340
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 0.6011462153583885,
|
| 2574 |
+
"grad_norm": 3.355443000793457,
|
| 2575 |
+
"learning_rate": 4.2009822440498684e-05,
|
| 2576 |
+
"loss": 0.8585092544555664,
|
| 2577 |
+
"step": 3350
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 0.6029406816728913,
|
| 2581 |
+
"grad_norm": 0.5484735369682312,
|
| 2582 |
+
"learning_rate": 4.1820929353985644e-05,
|
| 2583 |
+
"loss": 0.5951589107513428,
|
| 2584 |
+
"step": 3360
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.6047351479873939,
|
| 2588 |
+
"grad_norm": 0.603266179561615,
|
| 2589 |
+
"learning_rate": 4.163203626747261e-05,
|
| 2590 |
+
"loss": 0.5819008827209473,
|
| 2591 |
+
"step": 3370
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 0.6065296143018966,
|
| 2595 |
+
"grad_norm": 0.7201900482177734,
|
| 2596 |
+
"learning_rate": 4.1443143180959584e-05,
|
| 2597 |
+
"loss": 0.612189531326294,
|
| 2598 |
+
"step": 3380
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 0.6083240806163992,
|
| 2602 |
+
"grad_norm": 1.0490522384643555,
|
| 2603 |
+
"learning_rate": 4.1254250094446544e-05,
|
| 2604 |
+
"loss": 0.6810519695281982,
|
| 2605 |
+
"step": 3390
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 0.6101185469309018,
|
| 2609 |
+
"grad_norm": 2.729159116744995,
|
| 2610 |
+
"learning_rate": 4.106535700793351e-05,
|
| 2611 |
+
"loss": 0.7446739196777343,
|
| 2612 |
+
"step": 3400
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 0.6119130132454045,
|
| 2616 |
+
"grad_norm": 0.5799201726913452,
|
| 2617 |
+
"learning_rate": 4.087646392142048e-05,
|
| 2618 |
+
"loss": 0.5873437881469726,
|
| 2619 |
+
"step": 3410
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 0.6137074795599071,
|
| 2623 |
+
"grad_norm": 0.6728550791740417,
|
| 2624 |
+
"learning_rate": 4.0687570834907445e-05,
|
| 2625 |
+
"loss": 0.6205723762512207,
|
| 2626 |
+
"step": 3420
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 0.6155019458744098,
|
| 2630 |
+
"grad_norm": 0.7077367305755615,
|
| 2631 |
+
"learning_rate": 4.049867774839441e-05,
|
| 2632 |
+
"loss": 0.612656831741333,
|
| 2633 |
+
"step": 3430
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 0.6172964121889124,
|
| 2637 |
+
"grad_norm": 1.0263466835021973,
|
| 2638 |
+
"learning_rate": 4.030978466188138e-05,
|
| 2639 |
+
"loss": 0.6865530967712402,
|
| 2640 |
+
"step": 3440
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 0.619090878503415,
|
| 2644 |
+
"grad_norm": 2.840676784515381,
|
| 2645 |
+
"learning_rate": 4.012089157536834e-05,
|
| 2646 |
+
"loss": 0.7740938663482666,
|
| 2647 |
+
"step": 3450
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 0.6208853448179178,
|
| 2651 |
+
"grad_norm": 0.6163209080696106,
|
| 2652 |
+
"learning_rate": 3.993199848885531e-05,
|
| 2653 |
+
"loss": 0.6177750110626221,
|
| 2654 |
+
"step": 3460
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 0.6226798111324204,
|
| 2658 |
+
"grad_norm": 0.6922303438186646,
|
| 2659 |
+
"learning_rate": 3.974310540234228e-05,
|
| 2660 |
+
"loss": 0.5895265102386474,
|
| 2661 |
+
"step": 3470
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 0.6244742774469231,
|
| 2665 |
+
"grad_norm": 0.8489530086517334,
|
| 2666 |
+
"learning_rate": 3.955421231582924e-05,
|
| 2667 |
+
"loss": 0.5997925281524659,
|
| 2668 |
+
"step": 3480
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 0.6262687437614257,
|
| 2672 |
+
"grad_norm": 1.0187089443206787,
|
| 2673 |
+
"learning_rate": 3.936531922931621e-05,
|
| 2674 |
+
"loss": 0.6696516513824463,
|
| 2675 |
+
"step": 3490
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 0.6280632100759284,
|
| 2679 |
+
"grad_norm": 2.555328607559204,
|
| 2680 |
+
"learning_rate": 3.917642614280317e-05,
|
| 2681 |
+
"loss": 0.7553822994232178,
|
| 2682 |
+
"step": 3500
|
| 2683 |
}
|
| 2684 |
],
|
| 2685 |
"logging_steps": 10,
|
|
|
|
| 2699 |
"attributes": {}
|
| 2700 |
}
|
| 2701 |
},
|
| 2702 |
+
"total_flos": 7.086544521807368e+17,
|
| 2703 |
"train_batch_size": 2,
|
| 2704 |
"trial_name": null,
|
| 2705 |
"trial_params": null
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2000 → checkpoint-3500}/training_args.bin
RENAMED
|
File without changes
|
qwen3.5-2b-100pct-masked-weighted-v2/tb/events.out.tfevents.1790605476.02b48645612f.4580.0
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 sha256:7fa106c4ef8bd62d31856ff34d525c41f797a3215ac5afe8c74f482852feec1c
|
| 3 |
+
size 94103
|
qwen3.5-2b-100pct-masked-weighted-v2/test_eval_entity/metrics.json
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 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": 471,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
-
"premature_stop_rate": 0.
|
| 15 |
"n_examples": 1010
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.30927295823059103,
|
| 3 |
+
"wer_zwnj_normalized": 0.3049497763687407,
|
| 4 |
+
"exact_match": 0.005940594059405941,
|
| 5 |
+
"fix_rate": 0.26019976829382846,
|
| 6 |
+
"preservation_rate": 0.8012803603023967,
|
| 7 |
+
"targeted_score": 0.5307400642981126,
|
| 8 |
+
"fix_rate_lenient": 0.25113300492610835,
|
| 9 |
+
"preservation_rate_lenient": 0.8033434689452549,
|
| 10 |
+
"targeted_score_lenient": 0.5272382369356816,
|
| 11 |
+
"fix_rate_recoverable": 0.25451149137729745,
|
| 12 |
"unrecoverable_targets": 471,
|
| 13 |
+
"hallucination_rate": 0.0009900990099009901,
|
| 14 |
+
"premature_stop_rate": 0.100990099009901,
|
| 15 |
"n_examples": 1010
|
| 16 |
}
|
qwen3.5-2b-100pct-masked-weighted-v2/test_eval_entity/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
qwen3.5-2b-100pct-masked-weighted-v2/test_eval_typo/metrics.json
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 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": 68,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
-
"premature_stop_rate": 0.
|
| 15 |
"n_examples": 358
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.1789333946031503,
|
| 3 |
+
"wer_zwnj_normalized": 0.17290852926910213,
|
| 4 |
+
"exact_match": 0.01675977653631285,
|
| 5 |
+
"fix_rate": 0.3476153547886778,
|
| 6 |
+
"preservation_rate": 0.914209272925058,
|
| 7 |
+
"targeted_score": 0.630912313856868,
|
| 8 |
+
"fix_rate_lenient": 0.3382322500517491,
|
| 9 |
+
"preservation_rate_lenient": 0.9175481262269624,
|
| 10 |
+
"targeted_score_lenient": 0.6278901881393557,
|
| 11 |
+
"fix_rate_recoverable": 0.34264119252571906,
|
| 12 |
"unrecoverable_targets": 68,
|
| 13 |
+
"hallucination_rate": 0.008379888268156424,
|
| 14 |
+
"premature_stop_rate": 0.0446927374301676,
|
| 15 |
"n_examples": 358
|
| 16 |
}
|
qwen3.5-2b-100pct-masked-weighted-v2/test_eval_typo/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|