checkpoint-5000
Browse files- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/README.md +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/adapter_config.json +0 -0
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/adapter_model.safetensors +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/optimizer.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/rng_state.pth +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/scheduler.pt +1 -1
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/trainer_state.json +2208 -3
- qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/training_args.bin +0 -0
- qwen3.5-2b-50pct-masked-weighted/tb/events.out.tfevents.1790426135.b0aacb224392.4711.0 +2 -2
- qwen3.5-2b-50pct-masked-weighted/test_eval_entity/metrics.json +11 -11
- qwen3.5-2b-50pct-masked-weighted/test_eval_entity/predictions.jsonl +0 -0
- qwen3.5-2b-50pct-masked-weighted/test_eval_typo/metrics.json +11 -11
- qwen3.5-2b-50pct-masked-weighted/test_eval_typo/predictions.jsonl +0 -0
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/README.md
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/adapter_config.json
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/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 sha256:9dbdda4f3eecacbb01b14a68831a88d982953ea81a2898f3988e2e14c2895fc1
|
| 3 |
size 134604152
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/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:c7c85663939529b65581d6d0e5a345a132e93063e78b1dfd4a68555bf29536ee
|
| 3 |
size 269426431
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/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:da1a1a08da7dfc968b01be7324bc3ebfa442d6bd5180395216060ccf73d05715
|
| 3 |
size 14645
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/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:a2f67dfb2df28fc447f6646d2829c4f66fbd00443394432073ea821a202cd13a
|
| 3 |
size 1465
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/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,
|
|
@@ -1443,6 +1443,2211 @@
|
|
| 1443 |
"learning_rate": 8.639025311673594e-05,
|
| 1444 |
"loss": 1.0853967666625977,
|
| 1445 |
"step": 2000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1446 |
}
|
| 1447 |
],
|
| 1448 |
"logging_steps": 10,
|
|
@@ -1462,7 +3667,7 @@
|
|
| 1462 |
"attributes": {}
|
| 1463 |
}
|
| 1464 |
},
|
| 1465 |
-
"total_flos":
|
| 1466 |
"train_batch_size": 2,
|
| 1467 |
"trial_name": null,
|
| 1468 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.8973237319694013,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1443 |
"learning_rate": 8.639025311673594e-05,
|
| 1444 |
"loss": 1.0853967666625977,
|
| 1445 |
"step": 2000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.3589294927877605,
|
| 1449 |
+
"step": 2000,
|
| 1450 |
+
"test_entity_exact_match": 0.0049504950495049506,
|
| 1451 |
+
"test_entity_fix_rate": 0.27269311456930834,
|
| 1452 |
+
"test_entity_fix_rate_lenient": 0.26022988505747124,
|
| 1453 |
+
"test_entity_fix_rate_recoverable": 0.26397567414040835,
|
| 1454 |
+
"test_entity_hallucination_rate": 0.006930693069306931,
|
| 1455 |
+
"test_entity_preservation_rate": 0.8482998230657873,
|
| 1456 |
+
"test_entity_preservation_rate_lenient": 0.8518981820047984,
|
| 1457 |
+
"test_entity_targeted_score": 0.5604964688175478,
|
| 1458 |
+
"test_entity_targeted_score_lenient": 0.5560640335311349,
|
| 1459 |
+
"test_entity_wer": 0.27376415708628216,
|
| 1460 |
+
"test_entity_wer_zwnj_normalized": 0.2673380941706429
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.3589294927877605,
|
| 1464 |
+
"step": 2000,
|
| 1465 |
+
"test_typo_exact_match": 0.008379888268156424,
|
| 1466 |
+
"test_typo_fix_rate": 0.3557580457541683,
|
| 1467 |
+
"test_typo_fix_rate_lenient": 0.3446491409646036,
|
| 1468 |
+
"test_typo_fix_rate_recoverable": 0.3495166036149643,
|
| 1469 |
+
"test_typo_hallucination_rate": 0.008379888268156424,
|
| 1470 |
+
"test_typo_preservation_rate": 0.9167083174080498,
|
| 1471 |
+
"test_typo_preservation_rate_lenient": 0.919745670837118,
|
| 1472 |
+
"test_typo_targeted_score": 0.6362331815811091,
|
| 1473 |
+
"test_typo_targeted_score_lenient": 0.6321974059008608,
|
| 1474 |
+
"test_typo_wer": 0.172500063822726,
|
| 1475 |
+
"test_typo_wer_zwnj_normalized": 0.16473922034157923
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 0.3589294927877605,
|
| 1479 |
+
"step": 2000,
|
| 1480 |
+
"test_best_wer": 0.2231321104545041
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.3607241402516993,
|
| 1484 |
+
"grad_norm": 0.7390958666801453,
|
| 1485 |
+
"learning_rate": 8.62958065734794e-05,
|
| 1486 |
+
"loss": 0.7684628009796143,
|
| 1487 |
+
"step": 2010
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.3625187877156381,
|
| 1491 |
+
"grad_norm": 0.8893972039222717,
|
| 1492 |
+
"learning_rate": 8.62013600302229e-05,
|
| 1493 |
+
"loss": 0.7752983570098877,
|
| 1494 |
+
"step": 2020
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.3643134351795769,
|
| 1498 |
+
"grad_norm": 1.067131757736206,
|
| 1499 |
+
"learning_rate": 8.610691348696637e-05,
|
| 1500 |
+
"loss": 0.8518107414245606,
|
| 1501 |
+
"step": 2030
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.36610808264351574,
|
| 1505 |
+
"grad_norm": 1.5757921934127808,
|
| 1506 |
+
"learning_rate": 8.601246694370987e-05,
|
| 1507 |
+
"loss": 0.8482656478881836,
|
| 1508 |
+
"step": 2040
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.3679027301074545,
|
| 1512 |
+
"grad_norm": 3.651500701904297,
|
| 1513 |
+
"learning_rate": 8.591802040045335e-05,
|
| 1514 |
+
"loss": 1.0244155883789063,
|
| 1515 |
+
"step": 2050
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.3696973775713933,
|
| 1519 |
+
"grad_norm": 0.8204668760299683,
|
| 1520 |
+
"learning_rate": 8.582357385719682e-05,
|
| 1521 |
+
"loss": 0.8487360000610351,
|
| 1522 |
+
"step": 2060
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.37149202503533213,
|
| 1526 |
+
"grad_norm": 0.9089038372039795,
|
| 1527 |
+
"learning_rate": 8.572912731394032e-05,
|
| 1528 |
+
"loss": 0.8046296119689942,
|
| 1529 |
+
"step": 2070
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.37328667249927094,
|
| 1533 |
+
"grad_norm": 1.0486875772476196,
|
| 1534 |
+
"learning_rate": 8.563468077068379e-05,
|
| 1535 |
+
"loss": 0.7622116088867188,
|
| 1536 |
+
"step": 2080
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.3750813199632097,
|
| 1540 |
+
"grad_norm": 1.5056663751602173,
|
| 1541 |
+
"learning_rate": 8.554023422742729e-05,
|
| 1542 |
+
"loss": 0.8422880172729492,
|
| 1543 |
+
"step": 2090
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.3768759674271485,
|
| 1547 |
+
"grad_norm": 4.078341960906982,
|
| 1548 |
+
"learning_rate": 8.544578768417077e-05,
|
| 1549 |
+
"loss": 1.0107419967651368,
|
| 1550 |
+
"step": 2100
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.37867061489108733,
|
| 1554 |
+
"grad_norm": 0.8763484358787537,
|
| 1555 |
+
"learning_rate": 8.535134114091424e-05,
|
| 1556 |
+
"loss": 0.8312085151672364,
|
| 1557 |
+
"step": 2110
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.38046526235502615,
|
| 1561 |
+
"grad_norm": 0.982690155506134,
|
| 1562 |
+
"learning_rate": 8.525689459765774e-05,
|
| 1563 |
+
"loss": 0.7727385997772217,
|
| 1564 |
+
"step": 2120
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.38225990981896496,
|
| 1568 |
+
"grad_norm": 1.123055100440979,
|
| 1569 |
+
"learning_rate": 8.516244805440121e-05,
|
| 1570 |
+
"loss": 0.8653607368469238,
|
| 1571 |
+
"step": 2130
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.3840545572829037,
|
| 1575 |
+
"grad_norm": 1.557551622390747,
|
| 1576 |
+
"learning_rate": 8.50680015111447e-05,
|
| 1577 |
+
"loss": 0.934599781036377,
|
| 1578 |
+
"step": 2140
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.38584920474684253,
|
| 1582 |
+
"grad_norm": 3.7444405555725098,
|
| 1583 |
+
"learning_rate": 8.497355496788817e-05,
|
| 1584 |
+
"loss": 0.98599853515625,
|
| 1585 |
+
"step": 2150
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.38764385221078135,
|
| 1589 |
+
"grad_norm": 0.7720208168029785,
|
| 1590 |
+
"learning_rate": 8.487910842463166e-05,
|
| 1591 |
+
"loss": 0.7521912574768066,
|
| 1592 |
+
"step": 2160
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.38943849967472016,
|
| 1596 |
+
"grad_norm": 0.9204645752906799,
|
| 1597 |
+
"learning_rate": 8.478466188137515e-05,
|
| 1598 |
+
"loss": 0.8045943260192872,
|
| 1599 |
+
"step": 2170
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.391233147138659,
|
| 1603 |
+
"grad_norm": 1.1036076545715332,
|
| 1604 |
+
"learning_rate": 8.469021533811862e-05,
|
| 1605 |
+
"loss": 0.7903369426727295,
|
| 1606 |
+
"step": 2180
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.39302779460259774,
|
| 1610 |
+
"grad_norm": 1.4786419868469238,
|
| 1611 |
+
"learning_rate": 8.459576879486211e-05,
|
| 1612 |
+
"loss": 0.9231284141540528,
|
| 1613 |
+
"step": 2190
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.39482244206653655,
|
| 1617 |
+
"grad_norm": 3.327639579772949,
|
| 1618 |
+
"learning_rate": 8.450132225160559e-05,
|
| 1619 |
+
"loss": 0.9885027885437012,
|
| 1620 |
+
"step": 2200
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.39661708953047536,
|
| 1624 |
+
"grad_norm": 0.7114759683609009,
|
| 1625 |
+
"learning_rate": 8.440687570834907e-05,
|
| 1626 |
+
"loss": 0.7236516952514649,
|
| 1627 |
+
"step": 2210
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.3984117369944142,
|
| 1631 |
+
"grad_norm": 0.9788351058959961,
|
| 1632 |
+
"learning_rate": 8.431242916509257e-05,
|
| 1633 |
+
"loss": 0.8003618240356445,
|
| 1634 |
+
"step": 2220
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.40020638445835294,
|
| 1638 |
+
"grad_norm": 1.124056339263916,
|
| 1639 |
+
"learning_rate": 8.421798262183604e-05,
|
| 1640 |
+
"loss": 0.830434513092041,
|
| 1641 |
+
"step": 2230
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.40200103192229175,
|
| 1645 |
+
"grad_norm": 1.548984169960022,
|
| 1646 |
+
"learning_rate": 8.412353607857952e-05,
|
| 1647 |
+
"loss": 0.9224008560180664,
|
| 1648 |
+
"step": 2240
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.40379567938623057,
|
| 1652 |
+
"grad_norm": 3.2998907566070557,
|
| 1653 |
+
"learning_rate": 8.402908953532301e-05,
|
| 1654 |
+
"loss": 1.0086255073547363,
|
| 1655 |
+
"step": 2250
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.4055903268501694,
|
| 1659 |
+
"grad_norm": 0.8208487629890442,
|
| 1660 |
+
"learning_rate": 8.393464299206649e-05,
|
| 1661 |
+
"loss": 0.7369195938110351,
|
| 1662 |
+
"step": 2260
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.4073849743141082,
|
| 1666 |
+
"grad_norm": 0.8723757266998291,
|
| 1667 |
+
"learning_rate": 8.384019644880999e-05,
|
| 1668 |
+
"loss": 0.7349898338317871,
|
| 1669 |
+
"step": 2270
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.40917962177804695,
|
| 1673 |
+
"grad_norm": 1.0648058652877808,
|
| 1674 |
+
"learning_rate": 8.374574990555346e-05,
|
| 1675 |
+
"loss": 0.7378031253814697,
|
| 1676 |
+
"step": 2280
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.41097426924198577,
|
| 1680 |
+
"grad_norm": 1.4518014192581177,
|
| 1681 |
+
"learning_rate": 8.365130336229694e-05,
|
| 1682 |
+
"loss": 0.8286027908325195,
|
| 1683 |
+
"step": 2290
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.4127689167059246,
|
| 1687 |
+
"grad_norm": 3.4065778255462646,
|
| 1688 |
+
"learning_rate": 8.355685681904043e-05,
|
| 1689 |
+
"loss": 0.9797749519348145,
|
| 1690 |
+
"step": 2300
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.4145635641698634,
|
| 1694 |
+
"grad_norm": 0.7722201347351074,
|
| 1695 |
+
"learning_rate": 8.346241027578391e-05,
|
| 1696 |
+
"loss": 0.8020071029663086,
|
| 1697 |
+
"step": 2310
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.4163582116338022,
|
| 1701 |
+
"grad_norm": 0.963054358959198,
|
| 1702 |
+
"learning_rate": 8.336796373252739e-05,
|
| 1703 |
+
"loss": 0.7832077980041504,
|
| 1704 |
+
"step": 2320
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.41815285909774097,
|
| 1708 |
+
"grad_norm": 1.2207591533660889,
|
| 1709 |
+
"learning_rate": 8.327351718927088e-05,
|
| 1710 |
+
"loss": 0.8532069206237793,
|
| 1711 |
+
"step": 2330
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.4199475065616798,
|
| 1715 |
+
"grad_norm": 1.279175877571106,
|
| 1716 |
+
"learning_rate": 8.317907064601436e-05,
|
| 1717 |
+
"loss": 0.8882369041442871,
|
| 1718 |
+
"step": 2340
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.4217421540256186,
|
| 1722 |
+
"grad_norm": 3.7435832023620605,
|
| 1723 |
+
"learning_rate": 8.308462410275784e-05,
|
| 1724 |
+
"loss": 0.9548303604125976,
|
| 1725 |
+
"step": 2350
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.4235368014895574,
|
| 1729 |
+
"grad_norm": 0.8402184844017029,
|
| 1730 |
+
"learning_rate": 8.299017755950133e-05,
|
| 1731 |
+
"loss": 0.80802001953125,
|
| 1732 |
+
"step": 2360
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.42533144895349617,
|
| 1736 |
+
"grad_norm": 0.9928398728370667,
|
| 1737 |
+
"learning_rate": 8.289573101624481e-05,
|
| 1738 |
+
"loss": 0.7461267948150635,
|
| 1739 |
+
"step": 2370
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.427126096417435,
|
| 1743 |
+
"grad_norm": 1.029775857925415,
|
| 1744 |
+
"learning_rate": 8.280128447298829e-05,
|
| 1745 |
+
"loss": 0.7816200733184815,
|
| 1746 |
+
"step": 2380
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.4289207438813738,
|
| 1750 |
+
"grad_norm": 1.676102638244629,
|
| 1751 |
+
"learning_rate": 8.270683792973178e-05,
|
| 1752 |
+
"loss": 0.8625323295593261,
|
| 1753 |
+
"step": 2390
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.4307153913453126,
|
| 1757 |
+
"grad_norm": 4.379483699798584,
|
| 1758 |
+
"learning_rate": 8.261239138647526e-05,
|
| 1759 |
+
"loss": 1.0918493270874023,
|
| 1760 |
+
"step": 2400
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.43251003880925143,
|
| 1764 |
+
"grad_norm": 0.801175057888031,
|
| 1765 |
+
"learning_rate": 8.251794484321874e-05,
|
| 1766 |
+
"loss": 0.7958838462829589,
|
| 1767 |
+
"step": 2410
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.4343046862731902,
|
| 1771 |
+
"grad_norm": 0.9804407954216003,
|
| 1772 |
+
"learning_rate": 8.242349829996223e-05,
|
| 1773 |
+
"loss": 0.7912225723266602,
|
| 1774 |
+
"step": 2420
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.436099333737129,
|
| 1778 |
+
"grad_norm": 1.0723820924758911,
|
| 1779 |
+
"learning_rate": 8.232905175670571e-05,
|
| 1780 |
+
"loss": 0.7362414360046386,
|
| 1781 |
+
"step": 2430
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.4378939812010678,
|
| 1785 |
+
"grad_norm": 1.4297014474868774,
|
| 1786 |
+
"learning_rate": 8.223460521344919e-05,
|
| 1787 |
+
"loss": 0.8591861724853516,
|
| 1788 |
+
"step": 2440
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.43968862866500663,
|
| 1792 |
+
"grad_norm": 3.624446392059326,
|
| 1793 |
+
"learning_rate": 8.214015867019268e-05,
|
| 1794 |
+
"loss": 0.9524138450622559,
|
| 1795 |
+
"step": 2450
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.44148327612894545,
|
| 1799 |
+
"grad_norm": 0.805458664894104,
|
| 1800 |
+
"learning_rate": 8.204571212693616e-05,
|
| 1801 |
+
"loss": 0.7998561382293701,
|
| 1802 |
+
"step": 2460
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.4432779235928842,
|
| 1806 |
+
"grad_norm": 1.0265886783599854,
|
| 1807 |
+
"learning_rate": 8.195126558367964e-05,
|
| 1808 |
+
"loss": 0.7871425151824951,
|
| 1809 |
+
"step": 2470
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.445072571056823,
|
| 1813 |
+
"grad_norm": 1.1441900730133057,
|
| 1814 |
+
"learning_rate": 8.185681904042313e-05,
|
| 1815 |
+
"loss": 0.8011743545532226,
|
| 1816 |
+
"step": 2480
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.44686721852076183,
|
| 1820 |
+
"grad_norm": 1.324745774269104,
|
| 1821 |
+
"learning_rate": 8.176237249716661e-05,
|
| 1822 |
+
"loss": 0.8792869567871093,
|
| 1823 |
+
"step": 2490
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.44866186598470065,
|
| 1827 |
+
"grad_norm": 3.6218490600585938,
|
| 1828 |
+
"learning_rate": 8.16679259539101e-05,
|
| 1829 |
+
"loss": 1.0584125518798828,
|
| 1830 |
+
"step": 2500
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 0.4504565134486394,
|
| 1834 |
+
"grad_norm": 0.8318199515342712,
|
| 1835 |
+
"learning_rate": 8.157347941065358e-05,
|
| 1836 |
+
"loss": 0.8112331390380859,
|
| 1837 |
+
"step": 2510
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.4522511609125782,
|
| 1841 |
+
"grad_norm": 0.8700211644172668,
|
| 1842 |
+
"learning_rate": 8.147903286739706e-05,
|
| 1843 |
+
"loss": 0.7301114082336426,
|
| 1844 |
+
"step": 2520
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.45404580837651703,
|
| 1848 |
+
"grad_norm": 1.0621533393859863,
|
| 1849 |
+
"learning_rate": 8.138458632414054e-05,
|
| 1850 |
+
"loss": 0.7812650203704834,
|
| 1851 |
+
"step": 2530
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.45584045584045585,
|
| 1855 |
+
"grad_norm": 1.5809693336486816,
|
| 1856 |
+
"learning_rate": 8.129013978088403e-05,
|
| 1857 |
+
"loss": 0.9069185256958008,
|
| 1858 |
+
"step": 2540
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 0.45763510330439466,
|
| 1862 |
+
"grad_norm": 4.058928966522217,
|
| 1863 |
+
"learning_rate": 8.11956932376275e-05,
|
| 1864 |
+
"loss": 0.9913713455200195,
|
| 1865 |
+
"step": 2550
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.4594297507683334,
|
| 1869 |
+
"grad_norm": 0.8730033040046692,
|
| 1870 |
+
"learning_rate": 8.1101246694371e-05,
|
| 1871 |
+
"loss": 0.830630874633789,
|
| 1872 |
+
"step": 2560
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 0.46122439823227224,
|
| 1876 |
+
"grad_norm": 0.9506435990333557,
|
| 1877 |
+
"learning_rate": 8.100680015111446e-05,
|
| 1878 |
+
"loss": 0.7877472400665283,
|
| 1879 |
+
"step": 2570
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.46301904569621105,
|
| 1883 |
+
"grad_norm": 1.205980896949768,
|
| 1884 |
+
"learning_rate": 8.091235360785796e-05,
|
| 1885 |
+
"loss": 0.7747097015380859,
|
| 1886 |
+
"step": 2580
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.46481369316014987,
|
| 1890 |
+
"grad_norm": 1.4989122152328491,
|
| 1891 |
+
"learning_rate": 8.081790706460144e-05,
|
| 1892 |
+
"loss": 0.8825558662414551,
|
| 1893 |
+
"step": 2590
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.4666083406240887,
|
| 1897 |
+
"grad_norm": 3.2947559356689453,
|
| 1898 |
+
"learning_rate": 8.072346052134491e-05,
|
| 1899 |
+
"loss": 0.9716864585876465,
|
| 1900 |
+
"step": 2600
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.46840298808802744,
|
| 1904 |
+
"grad_norm": 0.8623685836791992,
|
| 1905 |
+
"learning_rate": 8.062901397808841e-05,
|
| 1906 |
+
"loss": 0.7903079032897949,
|
| 1907 |
+
"step": 2610
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 0.47019763555196625,
|
| 1911 |
+
"grad_norm": 0.8465585112571716,
|
| 1912 |
+
"learning_rate": 8.053456743483188e-05,
|
| 1913 |
+
"loss": 0.7284007549285889,
|
| 1914 |
+
"step": 2620
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.47199228301590507,
|
| 1918 |
+
"grad_norm": 1.1626179218292236,
|
| 1919 |
+
"learning_rate": 8.044012089157538e-05,
|
| 1920 |
+
"loss": 0.7963805198669434,
|
| 1921 |
+
"step": 2630
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.4737869304798439,
|
| 1925 |
+
"grad_norm": 1.5383787155151367,
|
| 1926 |
+
"learning_rate": 8.034567434831886e-05,
|
| 1927 |
+
"loss": 0.910552978515625,
|
| 1928 |
+
"step": 2640
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.47558157794378264,
|
| 1932 |
+
"grad_norm": 3.7534942626953125,
|
| 1933 |
+
"learning_rate": 8.025122780506233e-05,
|
| 1934 |
+
"loss": 1.0089344024658202,
|
| 1935 |
+
"step": 2650
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.47737622540772146,
|
| 1939 |
+
"grad_norm": 0.8134721517562866,
|
| 1940 |
+
"learning_rate": 8.015678126180583e-05,
|
| 1941 |
+
"loss": 0.8045218467712403,
|
| 1942 |
+
"step": 2660
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 0.47917087287166027,
|
| 1946 |
+
"grad_norm": 0.9830255508422852,
|
| 1947 |
+
"learning_rate": 8.00623347185493e-05,
|
| 1948 |
+
"loss": 0.8016125679016113,
|
| 1949 |
+
"step": 2670
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 0.4809655203355991,
|
| 1953 |
+
"grad_norm": 1.0700372457504272,
|
| 1954 |
+
"learning_rate": 7.99678881752928e-05,
|
| 1955 |
+
"loss": 0.9045357704162598,
|
| 1956 |
+
"step": 2680
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 0.4827601677995379,
|
| 1960 |
+
"grad_norm": 1.521283745765686,
|
| 1961 |
+
"learning_rate": 7.987344163203628e-05,
|
| 1962 |
+
"loss": 0.8992018699645996,
|
| 1963 |
+
"step": 2690
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 0.48455481526347666,
|
| 1967 |
+
"grad_norm": 3.5036544799804688,
|
| 1968 |
+
"learning_rate": 7.977899508877975e-05,
|
| 1969 |
+
"loss": 0.9581445693969727,
|
| 1970 |
+
"step": 2700
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.48634946272741547,
|
| 1974 |
+
"grad_norm": 0.8462426066398621,
|
| 1975 |
+
"learning_rate": 7.968454854552325e-05,
|
| 1976 |
+
"loss": 0.8029226303100586,
|
| 1977 |
+
"step": 2710
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.4881441101913543,
|
| 1981 |
+
"grad_norm": 1.0023834705352783,
|
| 1982 |
+
"learning_rate": 7.959010200226672e-05,
|
| 1983 |
+
"loss": 0.7849497318267822,
|
| 1984 |
+
"step": 2720
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 0.4899387576552931,
|
| 1988 |
+
"grad_norm": 1.1512731313705444,
|
| 1989 |
+
"learning_rate": 7.94956554590102e-05,
|
| 1990 |
+
"loss": 0.7798937320709228,
|
| 1991 |
+
"step": 2730
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 0.4917334051192319,
|
| 1995 |
+
"grad_norm": 1.4260687828063965,
|
| 1996 |
+
"learning_rate": 7.940120891575368e-05,
|
| 1997 |
+
"loss": 0.795078706741333,
|
| 1998 |
+
"step": 2740
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 0.4935280525831707,
|
| 2002 |
+
"grad_norm": 3.980698585510254,
|
| 2003 |
+
"learning_rate": 7.930676237249717e-05,
|
| 2004 |
+
"loss": 0.937680435180664,
|
| 2005 |
+
"step": 2750
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 0.4953227000471095,
|
| 2009 |
+
"grad_norm": 0.85591059923172,
|
| 2010 |
+
"learning_rate": 7.921231582924066e-05,
|
| 2011 |
+
"loss": 0.7800557136535644,
|
| 2012 |
+
"step": 2760
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.4971173475110483,
|
| 2016 |
+
"grad_norm": 0.8919742703437805,
|
| 2017 |
+
"learning_rate": 7.911786928598413e-05,
|
| 2018 |
+
"loss": 0.8073655128479004,
|
| 2019 |
+
"step": 2770
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.4989119949749871,
|
| 2023 |
+
"grad_norm": 1.1961230039596558,
|
| 2024 |
+
"learning_rate": 7.902342274272762e-05,
|
| 2025 |
+
"loss": 0.7749494075775146,
|
| 2026 |
+
"step": 2780
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.5007066424389259,
|
| 2030 |
+
"grad_norm": 1.384748101234436,
|
| 2031 |
+
"learning_rate": 7.89289761994711e-05,
|
| 2032 |
+
"loss": 0.8329132080078125,
|
| 2033 |
+
"step": 2790
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.5025012899028647,
|
| 2037 |
+
"grad_norm": 4.502509593963623,
|
| 2038 |
+
"learning_rate": 7.883452965621458e-05,
|
| 2039 |
+
"loss": 1.0134794235229492,
|
| 2040 |
+
"step": 2800
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 0.5042959373668036,
|
| 2044 |
+
"grad_norm": 0.7542349100112915,
|
| 2045 |
+
"learning_rate": 7.874008311295808e-05,
|
| 2046 |
+
"loss": 0.7716650009155274,
|
| 2047 |
+
"step": 2810
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.5060905848307423,
|
| 2051 |
+
"grad_norm": 0.8581972718238831,
|
| 2052 |
+
"learning_rate": 7.864563656970155e-05,
|
| 2053 |
+
"loss": 0.7554300785064697,
|
| 2054 |
+
"step": 2820
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.5078852322946811,
|
| 2058 |
+
"grad_norm": 1.0908201932907104,
|
| 2059 |
+
"learning_rate": 7.855119002644503e-05,
|
| 2060 |
+
"loss": 0.7669723033905029,
|
| 2061 |
+
"step": 2830
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.50967987975862,
|
| 2065 |
+
"grad_norm": 1.4047800302505493,
|
| 2066 |
+
"learning_rate": 7.845674348318852e-05,
|
| 2067 |
+
"loss": 0.8453067779541016,
|
| 2068 |
+
"step": 2840
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 0.5114745272225587,
|
| 2072 |
+
"grad_norm": 3.700296401977539,
|
| 2073 |
+
"learning_rate": 7.8362296939932e-05,
|
| 2074 |
+
"loss": 1.0116771697998046,
|
| 2075 |
+
"step": 2850
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 0.5132691746864976,
|
| 2079 |
+
"grad_norm": 0.8191012740135193,
|
| 2080 |
+
"learning_rate": 7.82678503966755e-05,
|
| 2081 |
+
"loss": 0.7772826671600341,
|
| 2082 |
+
"step": 2860
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 0.5150638221504363,
|
| 2086 |
+
"grad_norm": 0.855146050453186,
|
| 2087 |
+
"learning_rate": 7.817340385341897e-05,
|
| 2088 |
+
"loss": 0.7753990650177002,
|
| 2089 |
+
"step": 2870
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 0.5168584696143751,
|
| 2093 |
+
"grad_norm": 1.1639156341552734,
|
| 2094 |
+
"learning_rate": 7.807895731016245e-05,
|
| 2095 |
+
"loss": 0.7668899059295654,
|
| 2096 |
+
"step": 2880
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.518653117078314,
|
| 2100 |
+
"grad_norm": 1.4478026628494263,
|
| 2101 |
+
"learning_rate": 7.798451076690593e-05,
|
| 2102 |
+
"loss": 0.843500804901123,
|
| 2103 |
+
"step": 2890
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.5204477645422527,
|
| 2107 |
+
"grad_norm": 3.984922170639038,
|
| 2108 |
+
"learning_rate": 7.789006422364942e-05,
|
| 2109 |
+
"loss": 0.971616268157959,
|
| 2110 |
+
"step": 2900
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.5222424120061915,
|
| 2114 |
+
"grad_norm": 0.8589176535606384,
|
| 2115 |
+
"learning_rate": 7.77956176803929e-05,
|
| 2116 |
+
"loss": 0.7315890789031982,
|
| 2117 |
+
"step": 2910
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.5240370594701304,
|
| 2121 |
+
"grad_norm": 0.9845552444458008,
|
| 2122 |
+
"learning_rate": 7.770117113713638e-05,
|
| 2123 |
+
"loss": 0.7463840007781982,
|
| 2124 |
+
"step": 2920
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 0.5258317069340691,
|
| 2128 |
+
"grad_norm": 1.0363562107086182,
|
| 2129 |
+
"learning_rate": 7.760672459387987e-05,
|
| 2130 |
+
"loss": 0.8146579742431641,
|
| 2131 |
+
"step": 2930
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 0.527626354398008,
|
| 2135 |
+
"grad_norm": 1.4898813962936401,
|
| 2136 |
+
"learning_rate": 7.751227805062335e-05,
|
| 2137 |
+
"loss": 0.8084490776062012,
|
| 2138 |
+
"step": 2940
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.5294210018619467,
|
| 2142 |
+
"grad_norm": 3.68038272857666,
|
| 2143 |
+
"learning_rate": 7.741783150736683e-05,
|
| 2144 |
+
"loss": 0.912483024597168,
|
| 2145 |
+
"step": 2950
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.5312156493258855,
|
| 2149 |
+
"grad_norm": 0.76303631067276,
|
| 2150 |
+
"learning_rate": 7.732338496411032e-05,
|
| 2151 |
+
"loss": 0.694810438156128,
|
| 2152 |
+
"step": 2960
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.5330102967898244,
|
| 2156 |
+
"grad_norm": 0.9913522005081177,
|
| 2157 |
+
"learning_rate": 7.72289384208538e-05,
|
| 2158 |
+
"loss": 0.7986444473266602,
|
| 2159 |
+
"step": 2970
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 0.5348049442537631,
|
| 2163 |
+
"grad_norm": 1.2463011741638184,
|
| 2164 |
+
"learning_rate": 7.713449187759728e-05,
|
| 2165 |
+
"loss": 0.8069249153137207,
|
| 2166 |
+
"step": 2980
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 0.536599591717702,
|
| 2170 |
+
"grad_norm": 1.4866461753845215,
|
| 2171 |
+
"learning_rate": 7.704004533434077e-05,
|
| 2172 |
+
"loss": 0.8331960678100586,
|
| 2173 |
+
"step": 2990
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 0.5383942391816408,
|
| 2177 |
+
"grad_norm": 4.406412124633789,
|
| 2178 |
+
"learning_rate": 7.694559879108425e-05,
|
| 2179 |
+
"loss": 0.9865753173828125,
|
| 2180 |
+
"step": 3000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.5383942391816408,
|
| 2184 |
+
"step": 3000,
|
| 2185 |
+
"test_entity_exact_match": 0.0009900990099009901,
|
| 2186 |
+
"test_entity_fix_rate": 0.2468610076087297,
|
| 2187 |
+
"test_entity_fix_rate_lenient": 0.24013136288998357,
|
| 2188 |
+
"test_entity_fix_rate_recoverable": 0.2436261569819895,
|
| 2189 |
+
"test_entity_hallucination_rate": 0.12772277227722773,
|
| 2190 |
+
"test_entity_preservation_rate": 0.6939617178703554,
|
| 2191 |
+
"test_entity_preservation_rate_lenient": 0.6974968887520367,
|
| 2192 |
+
"test_entity_targeted_score": 0.47041136273954254,
|
| 2193 |
+
"test_entity_targeted_score_lenient": 0.4688141258210102,
|
| 2194 |
+
"test_entity_wer": 0.5446355184082151,
|
| 2195 |
+
"test_entity_wer_zwnj_normalized": 0.5383588988161954
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 0.5383942391816408,
|
| 2199 |
+
"step": 3000,
|
| 2200 |
+
"test_typo_exact_match": 0.002793296089385475,
|
| 2201 |
+
"test_typo_fix_rate": 0.33365645599069405,
|
| 2202 |
+
"test_typo_fix_rate_lenient": 0.32995239080935623,
|
| 2203 |
+
"test_typo_fix_rate_recoverable": 0.3343841950399327,
|
| 2204 |
+
"test_typo_hallucination_rate": 0.18156424581005587,
|
| 2205 |
+
"test_typo_preservation_rate": 0.7954311586746244,
|
| 2206 |
+
"test_typo_preservation_rate_lenient": 0.799349526795394,
|
| 2207 |
+
"test_typo_targeted_score": 0.5645438073326592,
|
| 2208 |
+
"test_typo_targeted_score_lenient": 0.5646509588023751,
|
| 2209 |
+
"test_typo_wer": 0.5629164432871256,
|
| 2210 |
+
"test_typo_wer_zwnj_normalized": 0.5535472671108729
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 0.5383942391816408,
|
| 2214 |
+
"step": 3000,
|
| 2215 |
+
"test_best_wer": 0.2231321104545041
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 0.5401888866455795,
|
| 2219 |
+
"grad_norm": 0.7907835245132446,
|
| 2220 |
+
"learning_rate": 7.685115224782773e-05,
|
| 2221 |
+
"loss": 0.7664157390594483,
|
| 2222 |
+
"step": 3010
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.5419835341095184,
|
| 2226 |
+
"grad_norm": 0.9218222498893738,
|
| 2227 |
+
"learning_rate": 7.675670570457122e-05,
|
| 2228 |
+
"loss": 0.6932620525360107,
|
| 2229 |
+
"step": 3020
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.5437781815734571,
|
| 2233 |
+
"grad_norm": 1.0903208255767822,
|
| 2234 |
+
"learning_rate": 7.66622591613147e-05,
|
| 2235 |
+
"loss": 0.714443302154541,
|
| 2236 |
+
"step": 3030
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 0.545572829037396,
|
| 2240 |
+
"grad_norm": 1.5942481756210327,
|
| 2241 |
+
"learning_rate": 7.656781261805818e-05,
|
| 2242 |
+
"loss": 0.8341629981994629,
|
| 2243 |
+
"step": 3040
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 0.5473674765013348,
|
| 2247 |
+
"grad_norm": 4.375911712646484,
|
| 2248 |
+
"learning_rate": 7.647336607480167e-05,
|
| 2249 |
+
"loss": 1.0025527954101563,
|
| 2250 |
+
"step": 3050
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.5491621239652735,
|
| 2254 |
+
"grad_norm": 0.8018923401832581,
|
| 2255 |
+
"learning_rate": 7.637891953154515e-05,
|
| 2256 |
+
"loss": 0.7377686977386475,
|
| 2257 |
+
"step": 3060
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.5509567714292124,
|
| 2261 |
+
"grad_norm": 0.8786219358444214,
|
| 2262 |
+
"learning_rate": 7.628447298828864e-05,
|
| 2263 |
+
"loss": 0.7309036254882812,
|
| 2264 |
+
"step": 3070
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.5527514188931512,
|
| 2268 |
+
"grad_norm": 1.1047521829605103,
|
| 2269 |
+
"learning_rate": 7.619002644503212e-05,
|
| 2270 |
+
"loss": 0.8081924438476562,
|
| 2271 |
+
"step": 3080
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 0.55454606635709,
|
| 2275 |
+
"grad_norm": 1.7077223062515259,
|
| 2276 |
+
"learning_rate": 7.609557990177559e-05,
|
| 2277 |
+
"loss": 0.8322297096252441,
|
| 2278 |
+
"step": 3090
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.5563407138210288,
|
| 2282 |
+
"grad_norm": 2.7238352298736572,
|
| 2283 |
+
"learning_rate": 7.600113335851909e-05,
|
| 2284 |
+
"loss": 0.9799996376037597,
|
| 2285 |
+
"step": 3100
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 0.5581353612849675,
|
| 2289 |
+
"grad_norm": 0.867943286895752,
|
| 2290 |
+
"learning_rate": 7.590668681526257e-05,
|
| 2291 |
+
"loss": 0.7850025177001954,
|
| 2292 |
+
"step": 3110
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.5599300087489064,
|
| 2296 |
+
"grad_norm": 1.0038567781448364,
|
| 2297 |
+
"learning_rate": 7.581224027200605e-05,
|
| 2298 |
+
"loss": 0.7511962890625,
|
| 2299 |
+
"step": 3120
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 0.5617246562128452,
|
| 2303 |
+
"grad_norm": 1.1457417011260986,
|
| 2304 |
+
"learning_rate": 7.571779372874954e-05,
|
| 2305 |
+
"loss": 0.7605049133300781,
|
| 2306 |
+
"step": 3130
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.563519303676784,
|
| 2310 |
+
"grad_norm": 1.5344796180725098,
|
| 2311 |
+
"learning_rate": 7.5623347185493e-05,
|
| 2312 |
+
"loss": 0.8576217651367187,
|
| 2313 |
+
"step": 3140
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.5653139511407228,
|
| 2317 |
+
"grad_norm": 4.3801422119140625,
|
| 2318 |
+
"learning_rate": 7.55289006422365e-05,
|
| 2319 |
+
"loss": 1.025908851623535,
|
| 2320 |
+
"step": 3150
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 0.5671085986046616,
|
| 2324 |
+
"grad_norm": 0.8065171241760254,
|
| 2325 |
+
"learning_rate": 7.543445409897997e-05,
|
| 2326 |
+
"loss": 0.7597884178161621,
|
| 2327 |
+
"step": 3160
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.5689032460686004,
|
| 2331 |
+
"grad_norm": 0.8725234866142273,
|
| 2332 |
+
"learning_rate": 7.534000755572347e-05,
|
| 2333 |
+
"loss": 0.7571310520172119,
|
| 2334 |
+
"step": 3170
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.5706978935325392,
|
| 2338 |
+
"grad_norm": 1.1811846494674683,
|
| 2339 |
+
"learning_rate": 7.524556101246695e-05,
|
| 2340 |
+
"loss": 0.8152105331420898,
|
| 2341 |
+
"step": 3180
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 0.572492540996478,
|
| 2345 |
+
"grad_norm": 1.403980016708374,
|
| 2346 |
+
"learning_rate": 7.515111446921042e-05,
|
| 2347 |
+
"loss": 0.7652612209320069,
|
| 2348 |
+
"step": 3190
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.5742871884604168,
|
| 2352 |
+
"grad_norm": 4.157504081726074,
|
| 2353 |
+
"learning_rate": 7.505666792595392e-05,
|
| 2354 |
+
"loss": 1.0274846076965332,
|
| 2355 |
+
"step": 3200
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 0.5760818359243556,
|
| 2359 |
+
"grad_norm": 0.945754885673523,
|
| 2360 |
+
"learning_rate": 7.496222138269739e-05,
|
| 2361 |
+
"loss": 0.776948881149292,
|
| 2362 |
+
"step": 3210
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 0.5778764833882944,
|
| 2366 |
+
"grad_norm": 0.9533429741859436,
|
| 2367 |
+
"learning_rate": 7.486777483944089e-05,
|
| 2368 |
+
"loss": 0.7675546646118164,
|
| 2369 |
+
"step": 3220
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 0.5796711308522332,
|
| 2373 |
+
"grad_norm": 1.1486726999282837,
|
| 2374 |
+
"learning_rate": 7.477332829618437e-05,
|
| 2375 |
+
"loss": 0.7956206798553467,
|
| 2376 |
+
"step": 3230
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 0.581465778316172,
|
| 2380 |
+
"grad_norm": 1.6006696224212646,
|
| 2381 |
+
"learning_rate": 7.467888175292784e-05,
|
| 2382 |
+
"loss": 0.7861947536468505,
|
| 2383 |
+
"step": 3240
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 0.5832604257801108,
|
| 2387 |
+
"grad_norm": 3.8867952823638916,
|
| 2388 |
+
"learning_rate": 7.458443520967134e-05,
|
| 2389 |
+
"loss": 0.9292440414428711,
|
| 2390 |
+
"step": 3250
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.5850550732440496,
|
| 2394 |
+
"grad_norm": 0.738664984703064,
|
| 2395 |
+
"learning_rate": 7.44899886664148e-05,
|
| 2396 |
+
"loss": 0.7322124004364013,
|
| 2397 |
+
"step": 3260
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 0.5868497207079885,
|
| 2401 |
+
"grad_norm": 0.9238191843032837,
|
| 2402 |
+
"learning_rate": 7.439554212315829e-05,
|
| 2403 |
+
"loss": 0.7845471858978271,
|
| 2404 |
+
"step": 3270
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.5886443681719272,
|
| 2408 |
+
"grad_norm": 1.1484863758087158,
|
| 2409 |
+
"learning_rate": 7.430109557990179e-05,
|
| 2410 |
+
"loss": 0.7382680892944335,
|
| 2411 |
+
"step": 3280
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 0.590439015635866,
|
| 2415 |
+
"grad_norm": 1.4992003440856934,
|
| 2416 |
+
"learning_rate": 7.420664903664526e-05,
|
| 2417 |
+
"loss": 0.8386685371398925,
|
| 2418 |
+
"step": 3290
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 0.5922336630998049,
|
| 2422 |
+
"grad_norm": 3.009324550628662,
|
| 2423 |
+
"learning_rate": 7.411220249338875e-05,
|
| 2424 |
+
"loss": 0.7852167606353759,
|
| 2425 |
+
"step": 3300
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 0.5940283105637436,
|
| 2429 |
+
"grad_norm": 0.7642790675163269,
|
| 2430 |
+
"learning_rate": 7.401775595013222e-05,
|
| 2431 |
+
"loss": 0.7435347080230713,
|
| 2432 |
+
"step": 3310
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.5958229580276825,
|
| 2436 |
+
"grad_norm": 0.9375689029693604,
|
| 2437 |
+
"learning_rate": 7.392330940687571e-05,
|
| 2438 |
+
"loss": 0.7808713436126709,
|
| 2439 |
+
"step": 3320
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.5976176054916212,
|
| 2443 |
+
"grad_norm": 1.0263559818267822,
|
| 2444 |
+
"learning_rate": 7.382886286361919e-05,
|
| 2445 |
+
"loss": 0.7671473979949951,
|
| 2446 |
+
"step": 3330
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.59941225295556,
|
| 2450 |
+
"grad_norm": 1.5290082693099976,
|
| 2451 |
+
"learning_rate": 7.373441632036267e-05,
|
| 2452 |
+
"loss": 0.8393860816955566,
|
| 2453 |
+
"step": 3340
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 0.6012069004194989,
|
| 2457 |
+
"grad_norm": 3.64926815032959,
|
| 2458 |
+
"learning_rate": 7.363996977710617e-05,
|
| 2459 |
+
"loss": 1.0867633819580078,
|
| 2460 |
+
"step": 3350
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 0.6030015478834376,
|
| 2464 |
+
"grad_norm": 0.808864176273346,
|
| 2465 |
+
"learning_rate": 7.354552323384964e-05,
|
| 2466 |
+
"loss": 0.7091707706451416,
|
| 2467 |
+
"step": 3360
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 0.6047961953473765,
|
| 2471 |
+
"grad_norm": 0.9375695586204529,
|
| 2472 |
+
"learning_rate": 7.345107669059312e-05,
|
| 2473 |
+
"loss": 0.7503323554992676,
|
| 2474 |
+
"step": 3370
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.6065908428113153,
|
| 2478 |
+
"grad_norm": 1.1794304847717285,
|
| 2479 |
+
"learning_rate": 7.335663014733661e-05,
|
| 2480 |
+
"loss": 0.7147882461547852,
|
| 2481 |
+
"step": 3380
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.608385490275254,
|
| 2485 |
+
"grad_norm": 1.5392098426818848,
|
| 2486 |
+
"learning_rate": 7.326218360408009e-05,
|
| 2487 |
+
"loss": 0.7968433380126954,
|
| 2488 |
+
"step": 3390
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 0.6101801377391929,
|
| 2492 |
+
"grad_norm": 4.211322784423828,
|
| 2493 |
+
"learning_rate": 7.316773706082359e-05,
|
| 2494 |
+
"loss": 1.0250563621520996,
|
| 2495 |
+
"step": 3400
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 0.6119747852031316,
|
| 2499 |
+
"grad_norm": 0.8506379127502441,
|
| 2500 |
+
"learning_rate": 7.307329051756706e-05,
|
| 2501 |
+
"loss": 0.7442142009735108,
|
| 2502 |
+
"step": 3410
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 0.6137694326670705,
|
| 2506 |
+
"grad_norm": 0.8980507254600525,
|
| 2507 |
+
"learning_rate": 7.297884397431054e-05,
|
| 2508 |
+
"loss": 0.7057757377624512,
|
| 2509 |
+
"step": 3420
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 0.6155640801310093,
|
| 2513 |
+
"grad_norm": 1.1606661081314087,
|
| 2514 |
+
"learning_rate": 7.288439743105402e-05,
|
| 2515 |
+
"loss": 0.7168849468231201,
|
| 2516 |
+
"step": 3430
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.617358727594948,
|
| 2520 |
+
"grad_norm": 1.7294520139694214,
|
| 2521 |
+
"learning_rate": 7.278995088779751e-05,
|
| 2522 |
+
"loss": 0.829401969909668,
|
| 2523 |
+
"step": 3440
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 0.6191533750588869,
|
| 2527 |
+
"grad_norm": 4.571506500244141,
|
| 2528 |
+
"learning_rate": 7.269550434454099e-05,
|
| 2529 |
+
"loss": 0.8971552848815918,
|
| 2530 |
+
"step": 3450
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.6209480225228257,
|
| 2534 |
+
"grad_norm": 0.8465120792388916,
|
| 2535 |
+
"learning_rate": 7.260105780128448e-05,
|
| 2536 |
+
"loss": 0.7918114185333252,
|
| 2537 |
+
"step": 3460
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 0.6227426699867645,
|
| 2541 |
+
"grad_norm": 0.9524924755096436,
|
| 2542 |
+
"learning_rate": 7.250661125802796e-05,
|
| 2543 |
+
"loss": 0.7100203990936279,
|
| 2544 |
+
"step": 3470
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.6245373174507033,
|
| 2548 |
+
"grad_norm": 1.2286087274551392,
|
| 2549 |
+
"learning_rate": 7.241216471477144e-05,
|
| 2550 |
+
"loss": 0.7934963703155518,
|
| 2551 |
+
"step": 3480
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 0.626331964914642,
|
| 2555 |
+
"grad_norm": 1.525288701057434,
|
| 2556 |
+
"learning_rate": 7.231771817151493e-05,
|
| 2557 |
+
"loss": 0.8523421287536621,
|
| 2558 |
+
"step": 3490
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.6281266123785809,
|
| 2562 |
+
"grad_norm": 3.8124051094055176,
|
| 2563 |
+
"learning_rate": 7.222327162825841e-05,
|
| 2564 |
+
"loss": 0.8778561592102051,
|
| 2565 |
+
"step": 3500
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 0.6299212598425197,
|
| 2569 |
+
"grad_norm": 0.784967303276062,
|
| 2570 |
+
"learning_rate": 7.212882508500189e-05,
|
| 2571 |
+
"loss": 0.7515722751617432,
|
| 2572 |
+
"step": 3510
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.6317159073064584,
|
| 2576 |
+
"grad_norm": 0.9887149333953857,
|
| 2577 |
+
"learning_rate": 7.203437854174538e-05,
|
| 2578 |
+
"loss": 0.7394243240356445,
|
| 2579 |
+
"step": 3520
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 0.6335105547703973,
|
| 2583 |
+
"grad_norm": 1.0454784631729126,
|
| 2584 |
+
"learning_rate": 7.193993199848886e-05,
|
| 2585 |
+
"loss": 0.7185003757476807,
|
| 2586 |
+
"step": 3530
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.6353052022343361,
|
| 2590 |
+
"grad_norm": 1.3647416830062866,
|
| 2591 |
+
"learning_rate": 7.184548545523234e-05,
|
| 2592 |
+
"loss": 0.8064111709594727,
|
| 2593 |
+
"step": 3540
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 0.6370998496982749,
|
| 2597 |
+
"grad_norm": 4.639543533325195,
|
| 2598 |
+
"learning_rate": 7.175103891197583e-05,
|
| 2599 |
+
"loss": 1.0075956344604493,
|
| 2600 |
+
"step": 3550
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.6388944971622137,
|
| 2604 |
+
"grad_norm": 0.7738682627677917,
|
| 2605 |
+
"learning_rate": 7.165659236871931e-05,
|
| 2606 |
+
"loss": 0.7540982246398926,
|
| 2607 |
+
"step": 3560
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 0.6406891446261525,
|
| 2611 |
+
"grad_norm": 0.8665961623191833,
|
| 2612 |
+
"learning_rate": 7.156214582546279e-05,
|
| 2613 |
+
"loss": 0.729641342163086,
|
| 2614 |
+
"step": 3570
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.6424837920900913,
|
| 2618 |
+
"grad_norm": 1.299155831336975,
|
| 2619 |
+
"learning_rate": 7.146769928220628e-05,
|
| 2620 |
+
"loss": 0.829476261138916,
|
| 2621 |
+
"step": 3580
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 0.6442784395540301,
|
| 2625 |
+
"grad_norm": 1.4906305074691772,
|
| 2626 |
+
"learning_rate": 7.137325273894976e-05,
|
| 2627 |
+
"loss": 0.8294624328613281,
|
| 2628 |
+
"step": 3590
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 0.646073087017969,
|
| 2632 |
+
"grad_norm": 4.132716178894043,
|
| 2633 |
+
"learning_rate": 7.127880619569324e-05,
|
| 2634 |
+
"loss": 0.9922947883605957,
|
| 2635 |
+
"step": 3600
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 0.6478677344819077,
|
| 2639 |
+
"grad_norm": 0.7767786979675293,
|
| 2640 |
+
"learning_rate": 7.118435965243673e-05,
|
| 2641 |
+
"loss": 0.8112610816955567,
|
| 2642 |
+
"step": 3610
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.6496623819458465,
|
| 2646 |
+
"grad_norm": 0.9437559843063354,
|
| 2647 |
+
"learning_rate": 7.108991310918021e-05,
|
| 2648 |
+
"loss": 0.7408264636993408,
|
| 2649 |
+
"step": 3620
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.6514570294097853,
|
| 2653 |
+
"grad_norm": 1.0995409488677979,
|
| 2654 |
+
"learning_rate": 7.099546656592368e-05,
|
| 2655 |
+
"loss": 0.6897705554962158,
|
| 2656 |
+
"step": 3630
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.6532516768737241,
|
| 2660 |
+
"grad_norm": 1.4682518243789673,
|
| 2661 |
+
"learning_rate": 7.090102002266718e-05,
|
| 2662 |
+
"loss": 0.8766571998596191,
|
| 2663 |
+
"step": 3640
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 0.655046324337663,
|
| 2667 |
+
"grad_norm": 3.8546547889709473,
|
| 2668 |
+
"learning_rate": 7.080657347941066e-05,
|
| 2669 |
+
"loss": 1.0745360374450683,
|
| 2670 |
+
"step": 3650
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.6568409718016017,
|
| 2674 |
+
"grad_norm": 0.8107970356941223,
|
| 2675 |
+
"learning_rate": 7.071212693615414e-05,
|
| 2676 |
+
"loss": 0.7250616073608398,
|
| 2677 |
+
"step": 3660
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.6586356192655405,
|
| 2681 |
+
"grad_norm": 1.0033174753189087,
|
| 2682 |
+
"learning_rate": 7.061768039289763e-05,
|
| 2683 |
+
"loss": 0.7318480491638184,
|
| 2684 |
+
"step": 3670
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.6604302667294794,
|
| 2688 |
+
"grad_norm": 1.1052309274673462,
|
| 2689 |
+
"learning_rate": 7.05232338496411e-05,
|
| 2690 |
+
"loss": 0.6873965263366699,
|
| 2691 |
+
"step": 3680
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 0.6622249141934181,
|
| 2695 |
+
"grad_norm": 1.4791178703308105,
|
| 2696 |
+
"learning_rate": 7.04287873063846e-05,
|
| 2697 |
+
"loss": 0.825076675415039,
|
| 2698 |
+
"step": 3690
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.664019561657357,
|
| 2702 |
+
"grad_norm": 3.5650668144226074,
|
| 2703 |
+
"learning_rate": 7.033434076312806e-05,
|
| 2704 |
+
"loss": 0.946928882598877,
|
| 2705 |
+
"step": 3700
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.6658142091212957,
|
| 2709 |
+
"grad_norm": 0.8390607237815857,
|
| 2710 |
+
"learning_rate": 7.023989421987156e-05,
|
| 2711 |
+
"loss": 0.7335081577301026,
|
| 2712 |
+
"step": 3710
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.6676088565852345,
|
| 2716 |
+
"grad_norm": 1.0318807363510132,
|
| 2717 |
+
"learning_rate": 7.014544767661504e-05,
|
| 2718 |
+
"loss": 0.7376636028289795,
|
| 2719 |
+
"step": 3720
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.6694035040491734,
|
| 2723 |
+
"grad_norm": 1.2701818943023682,
|
| 2724 |
+
"learning_rate": 7.005100113335851e-05,
|
| 2725 |
+
"loss": 0.741237735748291,
|
| 2726 |
+
"step": 3730
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.6711981515131121,
|
| 2730 |
+
"grad_norm": 1.5756694078445435,
|
| 2731 |
+
"learning_rate": 6.995655459010201e-05,
|
| 2732 |
+
"loss": 0.8473384857177735,
|
| 2733 |
+
"step": 3740
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.672992798977051,
|
| 2737 |
+
"grad_norm": 3.7947230339050293,
|
| 2738 |
+
"learning_rate": 6.986210804684548e-05,
|
| 2739 |
+
"loss": 0.8996176719665527,
|
| 2740 |
+
"step": 3750
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.6747874464409898,
|
| 2744 |
+
"grad_norm": 0.7768403887748718,
|
| 2745 |
+
"learning_rate": 6.976766150358898e-05,
|
| 2746 |
+
"loss": 0.7349163055419922,
|
| 2747 |
+
"step": 3760
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.6765820939049285,
|
| 2751 |
+
"grad_norm": 1.0075407028198242,
|
| 2752 |
+
"learning_rate": 6.967321496033246e-05,
|
| 2753 |
+
"loss": 0.7181667804718017,
|
| 2754 |
+
"step": 3770
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.6783767413688674,
|
| 2758 |
+
"grad_norm": 1.1049768924713135,
|
| 2759 |
+
"learning_rate": 6.957876841707593e-05,
|
| 2760 |
+
"loss": 0.7353485107421875,
|
| 2761 |
+
"step": 3780
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.6801713888328061,
|
| 2765 |
+
"grad_norm": 1.455216407775879,
|
| 2766 |
+
"learning_rate": 6.948432187381943e-05,
|
| 2767 |
+
"loss": 0.8581386566162109,
|
| 2768 |
+
"step": 3790
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.6819660362967449,
|
| 2772 |
+
"grad_norm": 3.471524953842163,
|
| 2773 |
+
"learning_rate": 6.93898753305629e-05,
|
| 2774 |
+
"loss": 0.8986720085144043,
|
| 2775 |
+
"step": 3800
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.6837606837606838,
|
| 2779 |
+
"grad_norm": 0.788811445236206,
|
| 2780 |
+
"learning_rate": 6.92954287873064e-05,
|
| 2781 |
+
"loss": 0.7062864303588867,
|
| 2782 |
+
"step": 3810
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.6855553312246225,
|
| 2786 |
+
"grad_norm": 1.0214735269546509,
|
| 2787 |
+
"learning_rate": 6.920098224404988e-05,
|
| 2788 |
+
"loss": 0.7787455081939697,
|
| 2789 |
+
"step": 3820
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.6873499786885614,
|
| 2793 |
+
"grad_norm": 1.1044228076934814,
|
| 2794 |
+
"learning_rate": 6.910653570079335e-05,
|
| 2795 |
+
"loss": 0.7103426933288575,
|
| 2796 |
+
"step": 3830
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.6891446261525002,
|
| 2800 |
+
"grad_norm": 1.486441731452942,
|
| 2801 |
+
"learning_rate": 6.901208915753685e-05,
|
| 2802 |
+
"loss": 0.8558782577514649,
|
| 2803 |
+
"step": 3840
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 0.6909392736164389,
|
| 2807 |
+
"grad_norm": 4.211693286895752,
|
| 2808 |
+
"learning_rate": 6.891764261428031e-05,
|
| 2809 |
+
"loss": 0.9754589080810547,
|
| 2810 |
+
"step": 3850
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.6927339210803778,
|
| 2814 |
+
"grad_norm": 0.7651219964027405,
|
| 2815 |
+
"learning_rate": 6.88231960710238e-05,
|
| 2816 |
+
"loss": 0.673793363571167,
|
| 2817 |
+
"step": 3860
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.6945285685443165,
|
| 2821 |
+
"grad_norm": 0.9120017886161804,
|
| 2822 |
+
"learning_rate": 6.87287495277673e-05,
|
| 2823 |
+
"loss": 0.6949755191802979,
|
| 2824 |
+
"step": 3870
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.6963232160082554,
|
| 2828 |
+
"grad_norm": 1.1558499336242676,
|
| 2829 |
+
"learning_rate": 6.863430298451077e-05,
|
| 2830 |
+
"loss": 0.7398390769958496,
|
| 2831 |
+
"step": 3880
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 0.6981178634721942,
|
| 2835 |
+
"grad_norm": 1.3961721658706665,
|
| 2836 |
+
"learning_rate": 6.853985644125426e-05,
|
| 2837 |
+
"loss": 0.8797501564025879,
|
| 2838 |
+
"step": 3890
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.6999125109361329,
|
| 2842 |
+
"grad_norm": 4.877803325653076,
|
| 2843 |
+
"learning_rate": 6.844540989799773e-05,
|
| 2844 |
+
"loss": 0.9983924865722656,
|
| 2845 |
+
"step": 3900
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 0.7017071584000718,
|
| 2849 |
+
"grad_norm": 0.7820862531661987,
|
| 2850 |
+
"learning_rate": 6.835096335474122e-05,
|
| 2851 |
+
"loss": 0.7368104934692383,
|
| 2852 |
+
"step": 3910
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.7035018058640106,
|
| 2856 |
+
"grad_norm": 0.8871352076530457,
|
| 2857 |
+
"learning_rate": 6.82565168114847e-05,
|
| 2858 |
+
"loss": 0.6663483142852783,
|
| 2859 |
+
"step": 3920
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.7052964533279494,
|
| 2863 |
+
"grad_norm": 1.1577374935150146,
|
| 2864 |
+
"learning_rate": 6.816207026822818e-05,
|
| 2865 |
+
"loss": 0.7413091659545898,
|
| 2866 |
+
"step": 3930
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.7070911007918882,
|
| 2870 |
+
"grad_norm": 1.3942395448684692,
|
| 2871 |
+
"learning_rate": 6.806762372497168e-05,
|
| 2872 |
+
"loss": 0.84088134765625,
|
| 2873 |
+
"step": 3940
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 0.708885748255827,
|
| 2877 |
+
"grad_norm": 3.1324944496154785,
|
| 2878 |
+
"learning_rate": 6.797317718171515e-05,
|
| 2879 |
+
"loss": 0.8465492248535156,
|
| 2880 |
+
"step": 3950
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.7106803957197658,
|
| 2884 |
+
"grad_norm": 0.7963802814483643,
|
| 2885 |
+
"learning_rate": 6.787873063845863e-05,
|
| 2886 |
+
"loss": 0.7199924468994141,
|
| 2887 |
+
"step": 3960
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.7124750431837046,
|
| 2891 |
+
"grad_norm": 0.9776906967163086,
|
| 2892 |
+
"learning_rate": 6.778428409520212e-05,
|
| 2893 |
+
"loss": 0.6692405700683594,
|
| 2894 |
+
"step": 3970
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.7142696906476435,
|
| 2898 |
+
"grad_norm": 1.1714221239089966,
|
| 2899 |
+
"learning_rate": 6.76898375519456e-05,
|
| 2900 |
+
"loss": 0.7462035179138183,
|
| 2901 |
+
"step": 3980
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.7160643381115822,
|
| 2905 |
+
"grad_norm": 1.5628330707550049,
|
| 2906 |
+
"learning_rate": 6.75953910086891e-05,
|
| 2907 |
+
"loss": 0.8628827095031738,
|
| 2908 |
+
"step": 3990
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.717858985575521,
|
| 2912 |
+
"grad_norm": 3.96386456489563,
|
| 2913 |
+
"learning_rate": 6.750094446543257e-05,
|
| 2914 |
+
"loss": 0.9594356536865234,
|
| 2915 |
+
"step": 4000
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.717858985575521,
|
| 2919 |
+
"step": 4000,
|
| 2920 |
+
"test_entity_exact_match": 0.0,
|
| 2921 |
+
"test_entity_fix_rate": 0.23646554153489682,
|
| 2922 |
+
"test_entity_fix_rate_lenient": 0.22922824302134648,
|
| 2923 |
+
"test_entity_fix_rate_recoverable": 0.23269956895111438,
|
| 2924 |
+
"test_entity_hallucination_rate": 0.13267326732673268,
|
| 2925 |
+
"test_entity_preservation_rate": 0.6708058549139456,
|
| 2926 |
+
"test_entity_preservation_rate_lenient": 0.6731650051961049,
|
| 2927 |
+
"test_entity_targeted_score": 0.45363569822442124,
|
| 2928 |
+
"test_entity_targeted_score_lenient": 0.4511966241087257,
|
| 2929 |
+
"test_entity_wer": 0.5756076472283601,
|
| 2930 |
+
"test_entity_wer_zwnj_normalized": 0.5715032930903812
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.717858985575521,
|
| 2934 |
+
"step": 4000,
|
| 2935 |
+
"test_typo_exact_match": 0.0,
|
| 2936 |
+
"test_typo_fix_rate": 0.31737107405971304,
|
| 2937 |
+
"test_typo_fix_rate_lenient": 0.31215069343821156,
|
| 2938 |
+
"test_typo_fix_rate_recoverable": 0.316099201345103,
|
| 2939 |
+
"test_typo_hallucination_rate": 0.17039106145251395,
|
| 2940 |
+
"test_typo_preservation_rate": 0.786258195395878,
|
| 2941 |
+
"test_typo_preservation_rate_lenient": 0.7890943186146678,
|
| 2942 |
+
"test_typo_targeted_score": 0.5518146347277956,
|
| 2943 |
+
"test_typo_targeted_score_lenient": 0.5506225060264397,
|
| 2944 |
+
"test_typo_wer": 0.5879860100584616,
|
| 2945 |
+
"test_typo_wer_zwnj_normalized": 0.5820122029052105
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 0.717858985575521,
|
| 2949 |
+
"step": 4000,
|
| 2950 |
+
"test_best_wer": 0.2231321104545041
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.7196536330394598,
|
| 2954 |
+
"grad_norm": 0.8268827795982361,
|
| 2955 |
+
"learning_rate": 6.740649792217605e-05,
|
| 2956 |
+
"loss": 0.7053637981414795,
|
| 2957 |
+
"step": 4010
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.7214482805033986,
|
| 2961 |
+
"grad_norm": 1.0276139974594116,
|
| 2962 |
+
"learning_rate": 6.731205137891953e-05,
|
| 2963 |
+
"loss": 0.7204613208770752,
|
| 2964 |
+
"step": 4020
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.7232429279673375,
|
| 2968 |
+
"grad_norm": 1.1466572284698486,
|
| 2969 |
+
"learning_rate": 6.721760483566302e-05,
|
| 2970 |
+
"loss": 0.6874489784240723,
|
| 2971 |
+
"step": 4030
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.7250375754312762,
|
| 2975 |
+
"grad_norm": 1.6566600799560547,
|
| 2976 |
+
"learning_rate": 6.71231582924065e-05,
|
| 2977 |
+
"loss": 0.8332732200622559,
|
| 2978 |
+
"step": 4040
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.726832222895215,
|
| 2982 |
+
"grad_norm": 3.702282190322876,
|
| 2983 |
+
"learning_rate": 6.702871174914998e-05,
|
| 2984 |
+
"loss": 0.8752366065979004,
|
| 2985 |
+
"step": 4050
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.7286268703591539,
|
| 2989 |
+
"grad_norm": 0.7897398471832275,
|
| 2990 |
+
"learning_rate": 6.693426520589347e-05,
|
| 2991 |
+
"loss": 0.7186062335968018,
|
| 2992 |
+
"step": 4060
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.7304215178230926,
|
| 2996 |
+
"grad_norm": 0.9881529211997986,
|
| 2997 |
+
"learning_rate": 6.683981866263695e-05,
|
| 2998 |
+
"loss": 0.731151008605957,
|
| 2999 |
+
"step": 4070
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.7322161652870315,
|
| 3003 |
+
"grad_norm": 1.118485927581787,
|
| 3004 |
+
"learning_rate": 6.674537211938043e-05,
|
| 3005 |
+
"loss": 0.7113176822662354,
|
| 3006 |
+
"step": 4080
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.7340108127509702,
|
| 3010 |
+
"grad_norm": 1.6644984483718872,
|
| 3011 |
+
"learning_rate": 6.665092557612392e-05,
|
| 3012 |
+
"loss": 0.8222126007080078,
|
| 3013 |
+
"step": 4090
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 0.735805460214909,
|
| 3017 |
+
"grad_norm": 2.4951610565185547,
|
| 3018 |
+
"learning_rate": 6.65564790328674e-05,
|
| 3019 |
+
"loss": 0.9295941352844238,
|
| 3020 |
+
"step": 4100
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.7376001076788479,
|
| 3024 |
+
"grad_norm": 0.8182881474494934,
|
| 3025 |
+
"learning_rate": 6.646203248961088e-05,
|
| 3026 |
+
"loss": 0.7256208419799804,
|
| 3027 |
+
"step": 4110
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 0.7393947551427866,
|
| 3031 |
+
"grad_norm": 0.9188361763954163,
|
| 3032 |
+
"learning_rate": 6.636758594635437e-05,
|
| 3033 |
+
"loss": 0.7089800357818603,
|
| 3034 |
+
"step": 4120
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.7411894026067254,
|
| 3038 |
+
"grad_norm": 1.1344740390777588,
|
| 3039 |
+
"learning_rate": 6.627313940309785e-05,
|
| 3040 |
+
"loss": 0.6811440467834473,
|
| 3041 |
+
"step": 4130
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.7429840500706643,
|
| 3045 |
+
"grad_norm": 1.3251533508300781,
|
| 3046 |
+
"learning_rate": 6.617869285984133e-05,
|
| 3047 |
+
"loss": 0.8110069274902344,
|
| 3048 |
+
"step": 4140
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.744778697534603,
|
| 3052 |
+
"grad_norm": 4.748185157775879,
|
| 3053 |
+
"learning_rate": 6.608424631658482e-05,
|
| 3054 |
+
"loss": 1.0361065864562988,
|
| 3055 |
+
"step": 4150
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.7465733449985419,
|
| 3059 |
+
"grad_norm": 0.8669788837432861,
|
| 3060 |
+
"learning_rate": 6.59897997733283e-05,
|
| 3061 |
+
"loss": 0.7604600429534912,
|
| 3062 |
+
"step": 4160
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.7483679924624806,
|
| 3066 |
+
"grad_norm": 0.8855018019676208,
|
| 3067 |
+
"learning_rate": 6.589535323007178e-05,
|
| 3068 |
+
"loss": 0.712589168548584,
|
| 3069 |
+
"step": 4170
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 0.7501626399264194,
|
| 3073 |
+
"grad_norm": 1.2251121997833252,
|
| 3074 |
+
"learning_rate": 6.580090668681527e-05,
|
| 3075 |
+
"loss": 0.7184146404266357,
|
| 3076 |
+
"step": 4180
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 0.7519572873903583,
|
| 3080 |
+
"grad_norm": 1.4125711917877197,
|
| 3081 |
+
"learning_rate": 6.570646014355875e-05,
|
| 3082 |
+
"loss": 0.8462773323059082,
|
| 3083 |
+
"step": 4190
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 0.753751934854297,
|
| 3087 |
+
"grad_norm": 3.009613275527954,
|
| 3088 |
+
"learning_rate": 6.561201360030223e-05,
|
| 3089 |
+
"loss": 0.9116678237915039,
|
| 3090 |
+
"step": 4200
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 0.7555465823182359,
|
| 3094 |
+
"grad_norm": 0.7657504081726074,
|
| 3095 |
+
"learning_rate": 6.551756705704572e-05,
|
| 3096 |
+
"loss": 0.7391473770141601,
|
| 3097 |
+
"step": 4210
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 0.7573412297821747,
|
| 3101 |
+
"grad_norm": 0.9948172569274902,
|
| 3102 |
+
"learning_rate": 6.542312051378919e-05,
|
| 3103 |
+
"loss": 0.7430819034576416,
|
| 3104 |
+
"step": 4220
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.7591358772461134,
|
| 3108 |
+
"grad_norm": 1.3323181867599487,
|
| 3109 |
+
"learning_rate": 6.532867397053269e-05,
|
| 3110 |
+
"loss": 0.7868825435638428,
|
| 3111 |
+
"step": 4230
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 0.7609305247100523,
|
| 3115 |
+
"grad_norm": 1.4949253797531128,
|
| 3116 |
+
"learning_rate": 6.523422742727617e-05,
|
| 3117 |
+
"loss": 0.8261573791503907,
|
| 3118 |
+
"step": 4240
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 0.762725172173991,
|
| 3122 |
+
"grad_norm": 4.123211860656738,
|
| 3123 |
+
"learning_rate": 6.513978088401965e-05,
|
| 3124 |
+
"loss": 0.9821381568908691,
|
| 3125 |
+
"step": 4250
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 0.7645198196379299,
|
| 3129 |
+
"grad_norm": 0.8264675140380859,
|
| 3130 |
+
"learning_rate": 6.504533434076314e-05,
|
| 3131 |
+
"loss": 0.7165445327758789,
|
| 3132 |
+
"step": 4260
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 0.7663144671018687,
|
| 3136 |
+
"grad_norm": 0.9300968050956726,
|
| 3137 |
+
"learning_rate": 6.49508877975066e-05,
|
| 3138 |
+
"loss": 0.7628948211669921,
|
| 3139 |
+
"step": 4270
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 0.7681091145658074,
|
| 3143 |
+
"grad_norm": 1.1202552318572998,
|
| 3144 |
+
"learning_rate": 6.48564412542501e-05,
|
| 3145 |
+
"loss": 0.7190794944763184,
|
| 3146 |
+
"step": 4280
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.7699037620297463,
|
| 3150 |
+
"grad_norm": 1.514113187789917,
|
| 3151 |
+
"learning_rate": 6.476199471099357e-05,
|
| 3152 |
+
"loss": 0.8148799896240234,
|
| 3153 |
+
"step": 4290
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 0.7716984094936851,
|
| 3157 |
+
"grad_norm": 3.2658019065856934,
|
| 3158 |
+
"learning_rate": 6.466754816773707e-05,
|
| 3159 |
+
"loss": 0.938233757019043,
|
| 3160 |
+
"step": 4300
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 0.7734930569576239,
|
| 3164 |
+
"grad_norm": 0.9332228302955627,
|
| 3165 |
+
"learning_rate": 6.457310162448055e-05,
|
| 3166 |
+
"loss": 0.7350199699401856,
|
| 3167 |
+
"step": 4310
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 0.7752877044215627,
|
| 3171 |
+
"grad_norm": 0.9837442636489868,
|
| 3172 |
+
"learning_rate": 6.447865508122402e-05,
|
| 3173 |
+
"loss": 0.7245872974395752,
|
| 3174 |
+
"step": 4320
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 0.7770823518855015,
|
| 3178 |
+
"grad_norm": 1.4489710330963135,
|
| 3179 |
+
"learning_rate": 6.438420853796752e-05,
|
| 3180 |
+
"loss": 0.7148647308349609,
|
| 3181 |
+
"step": 4330
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 0.7788769993494403,
|
| 3185 |
+
"grad_norm": 1.4325411319732666,
|
| 3186 |
+
"learning_rate": 6.428976199471099e-05,
|
| 3187 |
+
"loss": 0.7869903087615967,
|
| 3188 |
+
"step": 4340
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 0.7806716468133791,
|
| 3192 |
+
"grad_norm": 4.452845096588135,
|
| 3193 |
+
"learning_rate": 6.419531545145449e-05,
|
| 3194 |
+
"loss": 0.8885753631591797,
|
| 3195 |
+
"step": 4350
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 0.782466294277318,
|
| 3199 |
+
"grad_norm": 0.8960409760475159,
|
| 3200 |
+
"learning_rate": 6.410086890819797e-05,
|
| 3201 |
+
"loss": 0.692021369934082,
|
| 3202 |
+
"step": 4360
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 0.7842609417412567,
|
| 3206 |
+
"grad_norm": 0.9574803113937378,
|
| 3207 |
+
"learning_rate": 6.400642236494144e-05,
|
| 3208 |
+
"loss": 0.7156808376312256,
|
| 3209 |
+
"step": 4370
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 0.7860555892051955,
|
| 3213 |
+
"grad_norm": 1.0848815441131592,
|
| 3214 |
+
"learning_rate": 6.391197582168494e-05,
|
| 3215 |
+
"loss": 0.6753977775573731,
|
| 3216 |
+
"step": 4380
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 0.7878502366691343,
|
| 3220 |
+
"grad_norm": 1.3776320219039917,
|
| 3221 |
+
"learning_rate": 6.38175292784284e-05,
|
| 3222 |
+
"loss": 0.9061891555786132,
|
| 3223 |
+
"step": 4390
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 0.7896448841330731,
|
| 3227 |
+
"grad_norm": 3.7691376209259033,
|
| 3228 |
+
"learning_rate": 6.372308273517189e-05,
|
| 3229 |
+
"loss": 0.9163275718688965,
|
| 3230 |
+
"step": 4400
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 0.7914395315970119,
|
| 3234 |
+
"grad_norm": 0.8445814847946167,
|
| 3235 |
+
"learning_rate": 6.362863619191539e-05,
|
| 3236 |
+
"loss": 0.73052978515625,
|
| 3237 |
+
"step": 4410
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 0.7932341790609507,
|
| 3241 |
+
"grad_norm": 0.9680049419403076,
|
| 3242 |
+
"learning_rate": 6.353418964865886e-05,
|
| 3243 |
+
"loss": 0.7528566837310791,
|
| 3244 |
+
"step": 4420
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 0.7950288265248895,
|
| 3248 |
+
"grad_norm": 1.2166266441345215,
|
| 3249 |
+
"learning_rate": 6.343974310540235e-05,
|
| 3250 |
+
"loss": 0.7165881156921386,
|
| 3251 |
+
"step": 4430
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 0.7968234739888284,
|
| 3255 |
+
"grad_norm": 1.4934667348861694,
|
| 3256 |
+
"learning_rate": 6.334529656214582e-05,
|
| 3257 |
+
"loss": 0.7936708927154541,
|
| 3258 |
+
"step": 4440
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 0.7986181214527671,
|
| 3262 |
+
"grad_norm": 3.400346040725708,
|
| 3263 |
+
"learning_rate": 6.32508500188893e-05,
|
| 3264 |
+
"loss": 0.9394660949707031,
|
| 3265 |
+
"step": 4450
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 0.8004127689167059,
|
| 3269 |
+
"grad_norm": 0.8123710751533508,
|
| 3270 |
+
"learning_rate": 6.31564034756328e-05,
|
| 3271 |
+
"loss": 0.6932941436767578,
|
| 3272 |
+
"step": 4460
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 0.8022074163806447,
|
| 3276 |
+
"grad_norm": 0.986383318901062,
|
| 3277 |
+
"learning_rate": 6.306195693237627e-05,
|
| 3278 |
+
"loss": 0.7004058361053467,
|
| 3279 |
+
"step": 4470
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 0.8040020638445835,
|
| 3283 |
+
"grad_norm": 1.1018342971801758,
|
| 3284 |
+
"learning_rate": 6.296751038911977e-05,
|
| 3285 |
+
"loss": 0.7086675643920899,
|
| 3286 |
+
"step": 4480
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 0.8057967113085224,
|
| 3290 |
+
"grad_norm": 1.3698561191558838,
|
| 3291 |
+
"learning_rate": 6.287306384586324e-05,
|
| 3292 |
+
"loss": 0.8011689186096191,
|
| 3293 |
+
"step": 4490
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 0.8075913587724611,
|
| 3297 |
+
"grad_norm": 3.463945150375366,
|
| 3298 |
+
"learning_rate": 6.277861730260672e-05,
|
| 3299 |
+
"loss": 0.9159812927246094,
|
| 3300 |
+
"step": 4500
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 0.8093860062363999,
|
| 3304 |
+
"grad_norm": 0.8161361217498779,
|
| 3305 |
+
"learning_rate": 6.268417075935021e-05,
|
| 3306 |
+
"loss": 0.6879207134246826,
|
| 3307 |
+
"step": 4510
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 0.8111806537003388,
|
| 3311 |
+
"grad_norm": 1.03386390209198,
|
| 3312 |
+
"learning_rate": 6.258972421609369e-05,
|
| 3313 |
+
"loss": 0.7262889385223389,
|
| 3314 |
+
"step": 4520
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 0.8129753011642775,
|
| 3318 |
+
"grad_norm": 1.2165919542312622,
|
| 3319 |
+
"learning_rate": 6.249527767283719e-05,
|
| 3320 |
+
"loss": 0.7320633411407471,
|
| 3321 |
+
"step": 4530
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 0.8147699486282164,
|
| 3325 |
+
"grad_norm": 1.5375968217849731,
|
| 3326 |
+
"learning_rate": 6.240083112958066e-05,
|
| 3327 |
+
"loss": 0.7480690956115723,
|
| 3328 |
+
"step": 4540
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 0.8165645960921551,
|
| 3332 |
+
"grad_norm": 5.311518669128418,
|
| 3333 |
+
"learning_rate": 6.230638458632414e-05,
|
| 3334 |
+
"loss": 0.947059154510498,
|
| 3335 |
+
"step": 4550
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 0.8183592435560939,
|
| 3339 |
+
"grad_norm": 0.874262273311615,
|
| 3340 |
+
"learning_rate": 6.221193804306762e-05,
|
| 3341 |
+
"loss": 0.7604807853698731,
|
| 3342 |
+
"step": 4560
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 0.8201538910200328,
|
| 3346 |
+
"grad_norm": 0.949961245059967,
|
| 3347 |
+
"learning_rate": 6.211749149981111e-05,
|
| 3348 |
+
"loss": 0.715032958984375,
|
| 3349 |
+
"step": 4570
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 0.8219485384839715,
|
| 3353 |
+
"grad_norm": 1.1289318799972534,
|
| 3354 |
+
"learning_rate": 6.202304495655459e-05,
|
| 3355 |
+
"loss": 0.6810876846313476,
|
| 3356 |
+
"step": 4580
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 0.8237431859479104,
|
| 3360 |
+
"grad_norm": 1.4804030656814575,
|
| 3361 |
+
"learning_rate": 6.192859841329807e-05,
|
| 3362 |
+
"loss": 0.7983368873596192,
|
| 3363 |
+
"step": 4590
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 0.8255378334118492,
|
| 3367 |
+
"grad_norm": 4.038904666900635,
|
| 3368 |
+
"learning_rate": 6.183415187004156e-05,
|
| 3369 |
+
"loss": 0.9614683151245117,
|
| 3370 |
+
"step": 4600
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 0.8273324808757879,
|
| 3374 |
+
"grad_norm": 0.8533964157104492,
|
| 3375 |
+
"learning_rate": 6.173970532678504e-05,
|
| 3376 |
+
"loss": 0.6545410633087159,
|
| 3377 |
+
"step": 4610
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 0.8291271283397268,
|
| 3381 |
+
"grad_norm": 0.9574486017227173,
|
| 3382 |
+
"learning_rate": 6.164525878352852e-05,
|
| 3383 |
+
"loss": 0.6657931327819824,
|
| 3384 |
+
"step": 4620
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 0.8309217758036656,
|
| 3388 |
+
"grad_norm": 1.1819963455200195,
|
| 3389 |
+
"learning_rate": 6.155081224027201e-05,
|
| 3390 |
+
"loss": 0.7103557109832763,
|
| 3391 |
+
"step": 4630
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 0.8327164232676044,
|
| 3395 |
+
"grad_norm": 1.3845491409301758,
|
| 3396 |
+
"learning_rate": 6.145636569701549e-05,
|
| 3397 |
+
"loss": 0.7800614356994628,
|
| 3398 |
+
"step": 4640
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 0.8345110707315432,
|
| 3402 |
+
"grad_norm": 3.9027099609375,
|
| 3403 |
+
"learning_rate": 6.136191915375898e-05,
|
| 3404 |
+
"loss": 0.9763243675231934,
|
| 3405 |
+
"step": 4650
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 0.8363057181954819,
|
| 3409 |
+
"grad_norm": 0.8903900384902954,
|
| 3410 |
+
"learning_rate": 6.126747261050246e-05,
|
| 3411 |
+
"loss": 0.6765525817871094,
|
| 3412 |
+
"step": 4660
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 0.8381003656594208,
|
| 3416 |
+
"grad_norm": 1.0684949159622192,
|
| 3417 |
+
"learning_rate": 6.117302606724594e-05,
|
| 3418 |
+
"loss": 0.7025310516357421,
|
| 3419 |
+
"step": 4670
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 0.8398950131233596,
|
| 3423 |
+
"grad_norm": 1.1708343029022217,
|
| 3424 |
+
"learning_rate": 6.107857952398943e-05,
|
| 3425 |
+
"loss": 0.7327500820159912,
|
| 3426 |
+
"step": 4680
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 0.8416896605872983,
|
| 3430 |
+
"grad_norm": 1.548041820526123,
|
| 3431 |
+
"learning_rate": 6.098413298073291e-05,
|
| 3432 |
+
"loss": 0.8242672920227051,
|
| 3433 |
+
"step": 4690
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 0.8434843080512372,
|
| 3437 |
+
"grad_norm": 4.831801891326904,
|
| 3438 |
+
"learning_rate": 6.088968643747639e-05,
|
| 3439 |
+
"loss": 0.9189159393310546,
|
| 3440 |
+
"step": 4700
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 0.845278955515176,
|
| 3444 |
+
"grad_norm": 0.7827987670898438,
|
| 3445 |
+
"learning_rate": 6.079523989421988e-05,
|
| 3446 |
+
"loss": 0.6926255702972413,
|
| 3447 |
+
"step": 4710
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 0.8470736029791148,
|
| 3451 |
+
"grad_norm": 1.0675493478775024,
|
| 3452 |
+
"learning_rate": 6.070079335096336e-05,
|
| 3453 |
+
"loss": 0.6892650604248047,
|
| 3454 |
+
"step": 4720
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 0.8488682504430536,
|
| 3458 |
+
"grad_norm": 1.4141958951950073,
|
| 3459 |
+
"learning_rate": 6.0606346807706836e-05,
|
| 3460 |
+
"loss": 0.7153830528259277,
|
| 3461 |
+
"step": 4730
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 0.8506628979069923,
|
| 3465 |
+
"grad_norm": 1.5664619207382202,
|
| 3466 |
+
"learning_rate": 6.0511900264450326e-05,
|
| 3467 |
+
"loss": 0.8375289916992188,
|
| 3468 |
+
"step": 4740
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 0.8524575453709312,
|
| 3472 |
+
"grad_norm": 4.490540027618408,
|
| 3473 |
+
"learning_rate": 6.041745372119381e-05,
|
| 3474 |
+
"loss": 0.8840279579162598,
|
| 3475 |
+
"step": 4750
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 0.85425219283487,
|
| 3479 |
+
"grad_norm": 0.8381492495536804,
|
| 3480 |
+
"learning_rate": 6.0323007177937286e-05,
|
| 3481 |
+
"loss": 0.6521424293518067,
|
| 3482 |
+
"step": 4760
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 0.8560468402988088,
|
| 3486 |
+
"grad_norm": 0.9900016784667969,
|
| 3487 |
+
"learning_rate": 6.0228560634680776e-05,
|
| 3488 |
+
"loss": 0.7005383491516113,
|
| 3489 |
+
"step": 4770
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 0.8578414877627476,
|
| 3493 |
+
"grad_norm": 1.2890015840530396,
|
| 3494 |
+
"learning_rate": 6.013411409142425e-05,
|
| 3495 |
+
"loss": 0.6942355632781982,
|
| 3496 |
+
"step": 4780
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 0.8596361352266864,
|
| 3500 |
+
"grad_norm": 1.5667506456375122,
|
| 3501 |
+
"learning_rate": 6.003966754816774e-05,
|
| 3502 |
+
"loss": 0.798072099685669,
|
| 3503 |
+
"step": 4790
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 0.8614307826906252,
|
| 3507 |
+
"grad_norm": 3.8494532108306885,
|
| 3508 |
+
"learning_rate": 5.994522100491122e-05,
|
| 3509 |
+
"loss": 0.959134578704834,
|
| 3510 |
+
"step": 4800
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 0.863225430154564,
|
| 3514 |
+
"grad_norm": 0.8042945265769958,
|
| 3515 |
+
"learning_rate": 5.98507744616547e-05,
|
| 3516 |
+
"loss": 0.7243553638458252,
|
| 3517 |
+
"step": 4810
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 0.8650200776185029,
|
| 3521 |
+
"grad_norm": 0.9480759501457214,
|
| 3522 |
+
"learning_rate": 5.975632791839819e-05,
|
| 3523 |
+
"loss": 0.6505447864532471,
|
| 3524 |
+
"step": 4820
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 0.8668147250824416,
|
| 3528 |
+
"grad_norm": 1.2419155836105347,
|
| 3529 |
+
"learning_rate": 5.966188137514167e-05,
|
| 3530 |
+
"loss": 0.7647704124450684,
|
| 3531 |
+
"step": 4830
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 0.8686093725463804,
|
| 3535 |
+
"grad_norm": 1.612558364868164,
|
| 3536 |
+
"learning_rate": 5.956743483188516e-05,
|
| 3537 |
+
"loss": 0.8548855781555176,
|
| 3538 |
+
"step": 4840
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 0.8704040200103192,
|
| 3542 |
+
"grad_norm": 4.58271598815918,
|
| 3543 |
+
"learning_rate": 5.947298828862864e-05,
|
| 3544 |
+
"loss": 0.843088436126709,
|
| 3545 |
+
"step": 4850
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 0.872198667474258,
|
| 3549 |
+
"grad_norm": 0.8638464212417603,
|
| 3550 |
+
"learning_rate": 5.937854174537212e-05,
|
| 3551 |
+
"loss": 0.649378776550293,
|
| 3552 |
+
"step": 4860
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 0.8739933149381969,
|
| 3556 |
+
"grad_norm": 0.8889257311820984,
|
| 3557 |
+
"learning_rate": 5.928409520211561e-05,
|
| 3558 |
+
"loss": 0.7091721057891845,
|
| 3559 |
+
"step": 4870
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 0.8757879624021356,
|
| 3563 |
+
"grad_norm": 1.238863468170166,
|
| 3564 |
+
"learning_rate": 5.918964865885909e-05,
|
| 3565 |
+
"loss": 0.7717015743255615,
|
| 3566 |
+
"step": 4880
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 0.8775826098660744,
|
| 3570 |
+
"grad_norm": 1.5579394102096558,
|
| 3571 |
+
"learning_rate": 5.909520211560258e-05,
|
| 3572 |
+
"loss": 0.7820851325988769,
|
| 3573 |
+
"step": 4890
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 0.8793772573300133,
|
| 3577 |
+
"grad_norm": 3.1934468746185303,
|
| 3578 |
+
"learning_rate": 5.9000755572346054e-05,
|
| 3579 |
+
"loss": 0.849330997467041,
|
| 3580 |
+
"step": 4900
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 0.881171904793952,
|
| 3584 |
+
"grad_norm": 0.8095167875289917,
|
| 3585 |
+
"learning_rate": 5.890630902908954e-05,
|
| 3586 |
+
"loss": 0.729648208618164,
|
| 3587 |
+
"step": 4910
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 0.8829665522578909,
|
| 3591 |
+
"grad_norm": 1.033992052078247,
|
| 3592 |
+
"learning_rate": 5.881186248583303e-05,
|
| 3593 |
+
"loss": 0.6961370944976807,
|
| 3594 |
+
"step": 4920
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 0.8847611997218296,
|
| 3598 |
+
"grad_norm": 1.2645893096923828,
|
| 3599 |
+
"learning_rate": 5.8717415942576504e-05,
|
| 3600 |
+
"loss": 0.7022522926330567,
|
| 3601 |
+
"step": 4930
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 0.8865558471857684,
|
| 3605 |
+
"grad_norm": 1.6214739084243774,
|
| 3606 |
+
"learning_rate": 5.862296939931998e-05,
|
| 3607 |
+
"loss": 0.7820470809936524,
|
| 3608 |
+
"step": 4940
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 0.8883504946497073,
|
| 3612 |
+
"grad_norm": 3.837054967880249,
|
| 3613 |
+
"learning_rate": 5.852852285606347e-05,
|
| 3614 |
+
"loss": 0.8870712280273437,
|
| 3615 |
+
"step": 4950
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 0.890145142113646,
|
| 3619 |
+
"grad_norm": 0.9028179049491882,
|
| 3620 |
+
"learning_rate": 5.8434076312806954e-05,
|
| 3621 |
+
"loss": 0.7104959964752198,
|
| 3622 |
+
"step": 4960
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 0.8919397895775849,
|
| 3626 |
+
"grad_norm": 0.9380175471305847,
|
| 3627 |
+
"learning_rate": 5.833962976955044e-05,
|
| 3628 |
+
"loss": 0.7519885540008545,
|
| 3629 |
+
"step": 4970
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 0.8937344370415237,
|
| 3633 |
+
"grad_norm": 1.1569966077804565,
|
| 3634 |
+
"learning_rate": 5.824518322629392e-05,
|
| 3635 |
+
"loss": 0.7281805992126464,
|
| 3636 |
+
"step": 4980
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 0.8955290845054624,
|
| 3640 |
+
"grad_norm": 1.584986686706543,
|
| 3641 |
+
"learning_rate": 5.81507366830374e-05,
|
| 3642 |
+
"loss": 0.8723097801208496,
|
| 3643 |
+
"step": 4990
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 0.8973237319694013,
|
| 3647 |
+
"grad_norm": 4.8088274002075195,
|
| 3648 |
+
"learning_rate": 5.805629013978089e-05,
|
| 3649 |
+
"loss": 0.9866933822631836,
|
| 3650 |
+
"step": 5000
|
| 3651 |
}
|
| 3652 |
],
|
| 3653 |
"logging_steps": 10,
|
|
|
|
| 3667 |
"attributes": {}
|
| 3668 |
}
|
| 3669 |
},
|
| 3670 |
+
"total_flos": 5.092053469059487e+17,
|
| 3671 |
"train_batch_size": 2,
|
| 3672 |
"trial_name": null,
|
| 3673 |
"trial_params": null
|
qwen3.5-2b-50pct-masked-weighted/{checkpoint-2000 → checkpoint-5000}/training_args.bin
RENAMED
|
File without changes
|
qwen3.5-2b-50pct-masked-weighted/tb/events.out.tfevents.1790426135.b0aacb224392.4711.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6de3308c2c6b39f63f261017e00b22f84af5498964a8d31acde827d097131b7
|
| 3 |
+
size 121295
|
qwen3.5-2b-50pct-masked-weighted/test_eval_entity/metrics.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"wer": 0.
|
| 3 |
-
"wer_zwnj_normalized": 0.
|
| 4 |
-
"exact_match": 0.
|
| 5 |
-
"fix_rate": 0.
|
| 6 |
-
"preservation_rate": 0.
|
| 7 |
-
"targeted_score": 0.
|
| 8 |
-
"fix_rate_lenient": 0.
|
| 9 |
-
"preservation_rate_lenient": 0.
|
| 10 |
-
"targeted_score_lenient": 0.
|
| 11 |
-
"fix_rate_recoverable": 0.
|
| 12 |
"unrecoverable_targets": 523,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
"n_examples": 1010
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.5463007440142612,
|
| 3 |
+
"wer_zwnj_normalized": 0.5412837181498916,
|
| 4 |
+
"exact_match": 0.0009900990099009901,
|
| 5 |
+
"fix_rate": 0.2111344208911294,
|
| 6 |
+
"preservation_rate": 0.6017693421264275,
|
| 7 |
+
"targeted_score": 0.4064518815087785,
|
| 8 |
+
"fix_rate_lenient": 0.2051888341543514,
|
| 9 |
+
"preservation_rate_lenient": 0.604743209781507,
|
| 10 |
+
"targeted_score_lenient": 0.4049660219679292,
|
| 11 |
+
"fix_rate_recoverable": 0.2082400507902563,
|
| 12 |
"unrecoverable_targets": 523,
|
| 13 |
+
"hallucination_rate": 0.06732673267326733,
|
| 14 |
"n_examples": 1010
|
| 15 |
}
|
qwen3.5-2b-50pct-masked-weighted/test_eval_entity/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
qwen3.5-2b-50pct-masked-weighted/test_eval_typo/metrics.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"wer": 0.
|
| 3 |
-
"wer_zwnj_normalized": 0.
|
| 4 |
-
"exact_match": 0.
|
| 5 |
-
"fix_rate": 0.
|
| 6 |
-
"preservation_rate": 0.
|
| 7 |
-
"targeted_score": 0.
|
| 8 |
-
"fix_rate_lenient": 0.
|
| 9 |
-
"preservation_rate_lenient": 0.
|
| 10 |
-
"targeted_score_lenient": 0.
|
| 11 |
-
"fix_rate_recoverable": 0.
|
| 12 |
"unrecoverable_targets": 73,
|
| 13 |
-
"hallucination_rate": 0.
|
| 14 |
"n_examples": 358
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"wer": 0.5167343187562227,
|
| 3 |
+
"wer_zwnj_normalized": 0.5094329989022491,
|
| 4 |
+
"exact_match": 0.00558659217877095,
|
| 5 |
+
"fix_rate": 0.2919736331911594,
|
| 6 |
+
"preservation_rate": 0.7199894158116015,
|
| 7 |
+
"targeted_score": 0.5059815245013805,
|
| 8 |
+
"fix_rate_lenient": 0.2889670875595115,
|
| 9 |
+
"preservation_rate_lenient": 0.7239004951800522,
|
| 10 |
+
"targeted_score_lenient": 0.5064337913697818,
|
| 11 |
+
"fix_rate_recoverable": 0.29298024379991594,
|
| 12 |
"unrecoverable_targets": 73,
|
| 13 |
+
"hallucination_rate": 0.1005586592178771,
|
| 14 |
"n_examples": 358
|
| 15 |
}
|
qwen3.5-2b-50pct-masked-weighted/test_eval_typo/predictions.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|