checkpoint-4000
Browse files- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/README.md +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/adapter_config.json +0 -0
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/adapter_model.safetensors +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/optimizer.pt +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/rng_state.pth +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/scheduler.pt +1 -1
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/trainer_state.json +1164 -3
- qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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-2500 → checkpoint-4000}/README.md
RENAMED
|
File without changes
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/adapter_config.json
RENAMED
|
File without changes
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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 sha256:17f592ab0199f8e2545f602a7632becb2eacfc3656029725358a5359aa33a8e0
|
| 3 |
size 134604152
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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 sha256:1fd7b8e6aa17d208174d4849732412ee65ae5e61593e44c77851e96ccc649e78
|
| 3 |
size 269426431
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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 sha256:7ec6f7f36454d207195cba32f184fcf686a723016fa11949221655a27741b45e
|
| 3 |
size 14645
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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:a364336f5d699e1f36deb3769d44a1be8424a4ee91582bf2420cbc5bb02f521d
|
| 3 |
size 1465
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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,
|
|
@@ -1906,6 +1906,1167 @@
|
|
| 1906 |
"learning_rate": 5.806573479410654e-05,
|
| 1907 |
"loss": 0.7738859176635742,
|
| 1908 |
"step": 2500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1909 |
}
|
| 1910 |
],
|
| 1911 |
"logging_steps": 10,
|
|
@@ -1925,7 +3086,7 @@
|
|
| 1925 |
"attributes": {}
|
| 1926 |
}
|
| 1927 |
},
|
| 1928 |
-
"total_flos":
|
| 1929 |
"train_batch_size": 2,
|
| 1930 |
"trial_name": null,
|
| 1931 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.717786525801061,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 4000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 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 |
+
"epoch": 0.6280632100759284,
|
| 2686 |
+
"step": 3500,
|
| 2687 |
+
"test_entity_exact_match": 0.005940594059405941,
|
| 2688 |
+
"test_entity_fix_rate": 0.26019976829382846,
|
| 2689 |
+
"test_entity_fix_rate_lenient": 0.25113300492610835,
|
| 2690 |
+
"test_entity_fix_rate_recoverable": 0.25451149137729745,
|
| 2691 |
+
"test_entity_hallucination_rate": 0.0009900990099009901,
|
| 2692 |
+
"test_entity_premature_stop_rate": 0.100990099009901,
|
| 2693 |
+
"test_entity_preservation_rate": 0.8012803603023967,
|
| 2694 |
+
"test_entity_preservation_rate_lenient": 0.8033434689452549,
|
| 2695 |
+
"test_entity_targeted_score": 0.5307400642981126,
|
| 2696 |
+
"test_entity_targeted_score_lenient": 0.5272382369356816,
|
| 2697 |
+
"test_entity_wer": 0.30927295823059103,
|
| 2698 |
+
"test_entity_wer_zwnj_normalized": 0.3049497763687407
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.6280632100759284,
|
| 2702 |
+
"step": 3500,
|
| 2703 |
+
"test_typo_exact_match": 0.01675977653631285,
|
| 2704 |
+
"test_typo_fix_rate": 0.3476153547886778,
|
| 2705 |
+
"test_typo_fix_rate_lenient": 0.3382322500517491,
|
| 2706 |
+
"test_typo_fix_rate_recoverable": 0.34264119252571906,
|
| 2707 |
+
"test_typo_hallucination_rate": 0.008379888268156424,
|
| 2708 |
+
"test_typo_premature_stop_rate": 0.0446927374301676,
|
| 2709 |
+
"test_typo_preservation_rate": 0.914209272925058,
|
| 2710 |
+
"test_typo_preservation_rate_lenient": 0.9175481262269624,
|
| 2711 |
+
"test_typo_targeted_score": 0.630912313856868,
|
| 2712 |
+
"test_typo_targeted_score_lenient": 0.6278901881393557,
|
| 2713 |
+
"test_typo_wer": 0.1789333946031503,
|
| 2714 |
+
"test_typo_wer_zwnj_normalized": 0.17290852926910213
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.6280632100759284,
|
| 2718 |
+
"step": 3500,
|
| 2719 |
+
"test_best_wer": 0.1682596050185929
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.629857676390431,
|
| 2723 |
+
"grad_norm": 0.594057559967041,
|
| 2724 |
+
"learning_rate": 3.8987533056290146e-05,
|
| 2725 |
+
"loss": 0.6379923343658447,
|
| 2726 |
+
"step": 3510
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.6316521427049336,
|
| 2730 |
+
"grad_norm": 0.6532785892486572,
|
| 2731 |
+
"learning_rate": 3.8798639969777106e-05,
|
| 2732 |
+
"loss": 0.6092521667480468,
|
| 2733 |
+
"step": 3520
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.6334466090194363,
|
| 2737 |
+
"grad_norm": 0.8033592104911804,
|
| 2738 |
+
"learning_rate": 3.860974688326407e-05,
|
| 2739 |
+
"loss": 0.6203675270080566,
|
| 2740 |
+
"step": 3530
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.6352410753339389,
|
| 2744 |
+
"grad_norm": 1.118489146232605,
|
| 2745 |
+
"learning_rate": 3.842085379675104e-05,
|
| 2746 |
+
"loss": 0.7058998107910156,
|
| 2747 |
+
"step": 3540
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.6370355416484417,
|
| 2751 |
+
"grad_norm": 2.9393725395202637,
|
| 2752 |
+
"learning_rate": 3.823196071023801e-05,
|
| 2753 |
+
"loss": 0.7705009937286377,
|
| 2754 |
+
"step": 3550
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.6388300079629443,
|
| 2758 |
+
"grad_norm": 0.6190208196640015,
|
| 2759 |
+
"learning_rate": 3.8043067623724974e-05,
|
| 2760 |
+
"loss": 0.5795778751373291,
|
| 2761 |
+
"step": 3560
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.640624474277447,
|
| 2765 |
+
"grad_norm": 0.6888657808303833,
|
| 2766 |
+
"learning_rate": 3.785417453721194e-05,
|
| 2767 |
+
"loss": 0.5996123790740967,
|
| 2768 |
+
"step": 3570
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.6424189405919496,
|
| 2772 |
+
"grad_norm": 0.8217747807502747,
|
| 2773 |
+
"learning_rate": 3.766528145069891e-05,
|
| 2774 |
+
"loss": 0.6112364768981934,
|
| 2775 |
+
"step": 3580
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.6442134069064522,
|
| 2779 |
+
"grad_norm": 1.0920473337173462,
|
| 2780 |
+
"learning_rate": 3.7476388364185874e-05,
|
| 2781 |
+
"loss": 0.648440170288086,
|
| 2782 |
+
"step": 3590
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.6460078732209549,
|
| 2786 |
+
"grad_norm": 3.5313987731933594,
|
| 2787 |
+
"learning_rate": 3.728749527767284e-05,
|
| 2788 |
+
"loss": 0.808320140838623,
|
| 2789 |
+
"step": 3600
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.6478023395354575,
|
| 2793 |
+
"grad_norm": 0.540505051612854,
|
| 2794 |
+
"learning_rate": 3.70986021911598e-05,
|
| 2795 |
+
"loss": 0.6144362926483155,
|
| 2796 |
+
"step": 3610
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.6495968058499602,
|
| 2800 |
+
"grad_norm": 0.639191210269928,
|
| 2801 |
+
"learning_rate": 3.6909709104646775e-05,
|
| 2802 |
+
"loss": 0.6020387172698974,
|
| 2803 |
+
"step": 3620
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 0.6513912721644628,
|
| 2807 |
+
"grad_norm": 0.7896407842636108,
|
| 2808 |
+
"learning_rate": 3.6720816018133735e-05,
|
| 2809 |
+
"loss": 0.5654033660888672,
|
| 2810 |
+
"step": 3630
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.6531857384789654,
|
| 2814 |
+
"grad_norm": 1.0394206047058105,
|
| 2815 |
+
"learning_rate": 3.65319229316207e-05,
|
| 2816 |
+
"loss": 0.6442302703857422,
|
| 2817 |
+
"step": 3640
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.6549802047934682,
|
| 2821 |
+
"grad_norm": 2.4631218910217285,
|
| 2822 |
+
"learning_rate": 3.6343029845107675e-05,
|
| 2823 |
+
"loss": 0.737581205368042,
|
| 2824 |
+
"step": 3650
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.6567746711079708,
|
| 2828 |
+
"grad_norm": 0.6947884559631348,
|
| 2829 |
+
"learning_rate": 3.6154136758594635e-05,
|
| 2830 |
+
"loss": 0.6207414150238038,
|
| 2831 |
+
"step": 3660
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 0.6585691374224735,
|
| 2835 |
+
"grad_norm": 0.6546284556388855,
|
| 2836 |
+
"learning_rate": 3.59652436720816e-05,
|
| 2837 |
+
"loss": 0.5804379463195801,
|
| 2838 |
+
"step": 3670
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.6603636037369761,
|
| 2842 |
+
"grad_norm": 0.8040427565574646,
|
| 2843 |
+
"learning_rate": 3.577635058556857e-05,
|
| 2844 |
+
"loss": 0.6174962520599365,
|
| 2845 |
+
"step": 3680
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 0.6621580700514788,
|
| 2849 |
+
"grad_norm": 1.0611952543258667,
|
| 2850 |
+
"learning_rate": 3.5587457499055536e-05,
|
| 2851 |
+
"loss": 0.6572425842285157,
|
| 2852 |
+
"step": 3690
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.6639525363659814,
|
| 2856 |
+
"grad_norm": 3.129544734954834,
|
| 2857 |
+
"learning_rate": 3.53985644125425e-05,
|
| 2858 |
+
"loss": 0.7847175598144531,
|
| 2859 |
+
"step": 3700
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.665747002680484,
|
| 2863 |
+
"grad_norm": 0.6134805083274841,
|
| 2864 |
+
"learning_rate": 3.520967132602947e-05,
|
| 2865 |
+
"loss": 0.6172709941864014,
|
| 2866 |
+
"step": 3710
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.6675414689949867,
|
| 2870 |
+
"grad_norm": 0.6793216466903687,
|
| 2871 |
+
"learning_rate": 3.502077823951643e-05,
|
| 2872 |
+
"loss": 0.6093488693237304,
|
| 2873 |
+
"step": 3720
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 0.6693359353094893,
|
| 2877 |
+
"grad_norm": 0.8119182586669922,
|
| 2878 |
+
"learning_rate": 3.4831885153003403e-05,
|
| 2879 |
+
"loss": 0.5988525390625,
|
| 2880 |
+
"step": 3730
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.6711304016239921,
|
| 2884 |
+
"grad_norm": 1.0592997074127197,
|
| 2885 |
+
"learning_rate": 3.464299206649037e-05,
|
| 2886 |
+
"loss": 0.6723791599273682,
|
| 2887 |
+
"step": 3740
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.6729248679384947,
|
| 2891 |
+
"grad_norm": 3.1161043643951416,
|
| 2892 |
+
"learning_rate": 3.445409897997734e-05,
|
| 2893 |
+
"loss": 0.758587646484375,
|
| 2894 |
+
"step": 3750
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.6747193342529973,
|
| 2898 |
+
"grad_norm": 0.5930680632591248,
|
| 2899 |
+
"learning_rate": 3.4265205893464304e-05,
|
| 2900 |
+
"loss": 0.614569091796875,
|
| 2901 |
+
"step": 3760
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.6765138005675,
|
| 2905 |
+
"grad_norm": 0.7018576264381409,
|
| 2906 |
+
"learning_rate": 3.4076312806951264e-05,
|
| 2907 |
+
"loss": 0.5947975158691406,
|
| 2908 |
+
"step": 3770
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.6783082668820026,
|
| 2912 |
+
"grad_norm": 0.8689290881156921,
|
| 2913 |
+
"learning_rate": 3.388741972043824e-05,
|
| 2914 |
+
"loss": 0.6392403602600097,
|
| 2915 |
+
"step": 3780
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.6801027331965053,
|
| 2919 |
+
"grad_norm": 1.0972599983215332,
|
| 2920 |
+
"learning_rate": 3.36985266339252e-05,
|
| 2921 |
+
"loss": 0.6406892299652099,
|
| 2922 |
+
"step": 3790
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.6818971995110079,
|
| 2926 |
+
"grad_norm": 2.6399776935577393,
|
| 2927 |
+
"learning_rate": 3.3509633547412165e-05,
|
| 2928 |
+
"loss": 0.7668939113616944,
|
| 2929 |
+
"step": 3800
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 0.6836916658255106,
|
| 2933 |
+
"grad_norm": 0.5827666521072388,
|
| 2934 |
+
"learning_rate": 3.332074046089913e-05,
|
| 2935 |
+
"loss": 0.5843689918518067,
|
| 2936 |
+
"step": 3810
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.6854861321400132,
|
| 2940 |
+
"grad_norm": 0.6932247877120972,
|
| 2941 |
+
"learning_rate": 3.31318473743861e-05,
|
| 2942 |
+
"loss": 0.5733523845672608,
|
| 2943 |
+
"step": 3820
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 0.6872805984545158,
|
| 2947 |
+
"grad_norm": 0.8388057351112366,
|
| 2948 |
+
"learning_rate": 3.2942954287873065e-05,
|
| 2949 |
+
"loss": 0.6051725387573242,
|
| 2950 |
+
"step": 3830
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.6890750647690186,
|
| 2954 |
+
"grad_norm": 1.1730097532272339,
|
| 2955 |
+
"learning_rate": 3.275406120136003e-05,
|
| 2956 |
+
"loss": 0.6890604019165039,
|
| 2957 |
+
"step": 3840
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.6908695310835212,
|
| 2961 |
+
"grad_norm": 3.1862173080444336,
|
| 2962 |
+
"learning_rate": 3.2565168114847e-05,
|
| 2963 |
+
"loss": 0.7675668239593506,
|
| 2964 |
+
"step": 3850
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.6926639973980239,
|
| 2968 |
+
"grad_norm": 0.5261708498001099,
|
| 2969 |
+
"learning_rate": 3.2376275028333966e-05,
|
| 2970 |
+
"loss": 0.6003636360168457,
|
| 2971 |
+
"step": 3860
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.6944584637125265,
|
| 2975 |
+
"grad_norm": 0.7328461408615112,
|
| 2976 |
+
"learning_rate": 3.218738194182093e-05,
|
| 2977 |
+
"loss": 0.5707239627838134,
|
| 2978 |
+
"step": 3870
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.6962529300270291,
|
| 2982 |
+
"grad_norm": 0.8534484505653381,
|
| 2983 |
+
"learning_rate": 3.199848885530789e-05,
|
| 2984 |
+
"loss": 0.6307664394378663,
|
| 2985 |
+
"step": 3880
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.6980473963415318,
|
| 2989 |
+
"grad_norm": 1.1296844482421875,
|
| 2990 |
+
"learning_rate": 3.1809595768794866e-05,
|
| 2991 |
+
"loss": 0.6771279335021972,
|
| 2992 |
+
"step": 3890
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.6998418626560344,
|
| 2996 |
+
"grad_norm": 3.205347776412964,
|
| 2997 |
+
"learning_rate": 3.1620702682281826e-05,
|
| 2998 |
+
"loss": 0.7323418617248535,
|
| 2999 |
+
"step": 3900
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.7016363289705371,
|
| 3003 |
+
"grad_norm": 0.5762456059455872,
|
| 3004 |
+
"learning_rate": 3.143180959576879e-05,
|
| 3005 |
+
"loss": 0.5696917533874511,
|
| 3006 |
+
"step": 3910
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.7034307952850397,
|
| 3010 |
+
"grad_norm": 0.6364765763282776,
|
| 3011 |
+
"learning_rate": 3.124291650925577e-05,
|
| 3012 |
+
"loss": 0.5995479583740234,
|
| 3013 |
+
"step": 3920
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 0.7052252615995425,
|
| 3017 |
+
"grad_norm": 0.808103621006012,
|
| 3018 |
+
"learning_rate": 3.105402342274273e-05,
|
| 3019 |
+
"loss": 0.6079168796539307,
|
| 3020 |
+
"step": 3930
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.7070197279140451,
|
| 3024 |
+
"grad_norm": 1.0433599948883057,
|
| 3025 |
+
"learning_rate": 3.08651303362297e-05,
|
| 3026 |
+
"loss": 0.659568452835083,
|
| 3027 |
+
"step": 3940
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 0.7088141942285477,
|
| 3031 |
+
"grad_norm": 2.3059654235839844,
|
| 3032 |
+
"learning_rate": 3.067623724971666e-05,
|
| 3033 |
+
"loss": 0.7577475547790528,
|
| 3034 |
+
"step": 3950
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.7106086605430504,
|
| 3038 |
+
"grad_norm": 0.623231828212738,
|
| 3039 |
+
"learning_rate": 3.048734416320363e-05,
|
| 3040 |
+
"loss": 0.6336734294891357,
|
| 3041 |
+
"step": 3960
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.712403126857553,
|
| 3045 |
+
"grad_norm": 0.6546342372894287,
|
| 3046 |
+
"learning_rate": 3.0298451076690594e-05,
|
| 3047 |
+
"loss": 0.5807650566101075,
|
| 3048 |
+
"step": 3970
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.7141975931720557,
|
| 3052 |
+
"grad_norm": 0.7533226609230042,
|
| 3053 |
+
"learning_rate": 3.010955799017756e-05,
|
| 3054 |
+
"loss": 0.5835853576660156,
|
| 3055 |
+
"step": 3980
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.7159920594865583,
|
| 3059 |
+
"grad_norm": 1.0764480829238892,
|
| 3060 |
+
"learning_rate": 2.9920664903664524e-05,
|
| 3061 |
+
"loss": 0.659849214553833,
|
| 3062 |
+
"step": 3990
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.717786525801061,
|
| 3066 |
+
"grad_norm": 3.068281650543213,
|
| 3067 |
+
"learning_rate": 2.9731771817151495e-05,
|
| 3068 |
+
"loss": 0.7490633964538574,
|
| 3069 |
+
"step": 4000
|
| 3070 |
}
|
| 3071 |
],
|
| 3072 |
"logging_steps": 10,
|
|
|
|
| 3086 |
"attributes": {}
|
| 3087 |
}
|
| 3088 |
},
|
| 3089 |
+
"total_flos": 8.100580632944594e+17,
|
| 3090 |
"train_batch_size": 2,
|
| 3091 |
"trial_name": null,
|
| 3092 |
"trial_params": null
|
qwen3.5-2b-100pct-masked-weighted-v2/{checkpoint-2500 → checkpoint-4000}/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 sha256:7b56f7998aa62197f279915fce790127f9166eb886590bdcec19fd5e29861a94
|
| 3 |
+
size 106439
|
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": 0.2505310575250051,
|
| 3 |
+
"wer_zwnj_normalized": 0.24571684759983348,
|
| 4 |
+
"exact_match": 0.006930693069306931,
|
| 5 |
+
"fix_rate": 0.28474809781757837,
|
| 6 |
+
"preservation_rate": 0.8719575357889657,
|
| 7 |
+
"targeted_score": 0.5783528168032721,
|
| 8 |
+
"fix_rate_lenient": 0.2760262725779967,
|
| 9 |
+
"preservation_rate_lenient": 0.8733176808693532,
|
| 10 |
+
"targeted_score_lenient": 0.574671976723675,
|
| 11 |
+
"fix_rate_recoverable": 0.2796290736849128,
|
| 12 |
"unrecoverable_targets": 471,
|
| 13 |
+
"hallucination_rate": 0.0049504950495049506,
|
| 14 |
+
"premature_stop_rate": 0.03861386138613861,
|
| 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": 0.13221515917387863,
|
| 3 |
+
"wer_zwnj_normalized": 0.12601159020704092,
|
| 4 |
+
"exact_match": 0.019553072625698324,
|
| 5 |
+
"fix_rate": 0.36758433501357113,
|
| 6 |
+
"preservation_rate": 0.9593978772822156,
|
| 7 |
+
"targeted_score": 0.6634911061478934,
|
| 8 |
+
"fix_rate_lenient": 0.3605878700062099,
|
| 9 |
+
"preservation_rate_lenient": 0.96144041724047,
|
| 10 |
+
"targeted_score_lenient": 0.66101414362334,
|
| 11 |
+
"fix_rate_recoverable": 0.36552592903632164,
|
| 12 |
"unrecoverable_targets": 68,
|
| 13 |
+
"hallucination_rate": 0.00558659217877095,
|
| 14 |
+
"premature_stop_rate": 0.008379888268156424,
|
| 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
|
|
|