Add c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000010000 (milestone, cosine 2e-4 -> 5e-5) (INDEX.json, post-verification)
Browse files- INDEX.json +26 -1
INDEX.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"checkpoint_count":
|
| 3 |
"checkpoints": [
|
| 4 |
{
|
| 5 |
"repo_path": "c1-1/packed_temporal_grid/iter_000003625",
|
|
@@ -554,6 +554,31 @@
|
|
| 554 |
"grad_clip": true,
|
| 555 |
"clip_norm": 5.0
|
| 556 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 557 |
{
|
| 558 |
"repo_path": "c3-3/packed_temporal_grid/iter_000002000",
|
| 559 |
"model_family": "c3-3",
|
|
|
|
| 1 |
{
|
| 2 |
+
"checkpoint_count": 36,
|
| 3 |
"checkpoints": [
|
| 4 |
{
|
| 5 |
"repo_path": "c1-1/packed_temporal_grid/iter_000003625",
|
|
|
|
| 554 |
"grad_clip": true,
|
| 555 |
"clip_norm": 5.0
|
| 556 |
},
|
| 557 |
+
{
|
| 558 |
+
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000010000",
|
| 559 |
+
"model_family": "c3-3",
|
| 560 |
+
"memory_rope_mode": "source_aligned",
|
| 561 |
+
"optimizer_step": 10000,
|
| 562 |
+
"sample_exposure": 80000,
|
| 563 |
+
"wandb_run_id": "jhnefd4q",
|
| 564 |
+
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
|
| 565 |
+
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
|
| 566 |
+
"variant": "cosine_2e-4_to5e-5_clip5",
|
| 567 |
+
"lr_schedule": {
|
| 568 |
+
"cosine_min_factor": 0.25,
|
| 569 |
+
"cycle_length_steps": 10000,
|
| 570 |
+
"f_max": 1.0,
|
| 571 |
+
"f_start": 1e-06,
|
| 572 |
+
"final_lr": 5e-05,
|
| 573 |
+
"peak_lr": 0.0002,
|
| 574 |
+
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
|
| 575 |
+
"type": "cosine",
|
| 576 |
+
"warmup_steps": 100
|
| 577 |
+
},
|
| 578 |
+
"optimizer_lr": 0.0002,
|
| 579 |
+
"grad_clip": true,
|
| 580 |
+
"clip_norm": 5.0
|
| 581 |
+
},
|
| 582 |
{
|
| 583 |
"repo_path": "c3-3/packed_temporal_grid/iter_000002000",
|
| 584 |
"model_family": "c3-3",
|