rspec1: checkpoints exit
Browse files
runs/rspec1/best/lora_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f30ee092c4708740ba3dd8521e886d6326140d3d089b249837a5ef1b17b7b6c
|
| 3 |
+
size 147843609
|
runs/rspec1/best/twin_config.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "Qwen/Qwen2.5-Math-1.5B",
|
| 3 |
+
"bidirectional": false,
|
| 4 |
+
"teacher_style": "solution_after_raw",
|
| 5 |
+
"lora_rank": 32,
|
| 6 |
+
"lora_alpha": 32.0,
|
| 7 |
+
"step": 60,
|
| 8 |
+
"val_metrics": {
|
| 9 |
+
"ce_encoder": 0.12896888902288065,
|
| 10 |
+
"ce_natural": 0.17520572880456878,
|
| 11 |
+
"ce_delta": -0.04623683978168815,
|
| 12 |
+
"ce_delta_easy": -0.05129569332881163,
|
| 13 |
+
"ce_delta_medium": -0.046015213204959805,
|
| 14 |
+
"ce_delta_very_hard": -0.033353918823147595,
|
| 15 |
+
"ce_delta_hard": -0.04037966402842126
|
| 16 |
+
}
|
| 17 |
+
}
|
runs/rspec1/latest/lora_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0370c0016613cf103ad263363080c3056ec93c68a5645508dc3fd876c1ac86bc
|
| 3 |
+
size 147843609
|
runs/rspec1/latest/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:905abbe0d36d265c8985bb0df76b8d61646aa3a670bd0a477d7832d6c5766c57
|
| 3 |
+
size 286123933
|
runs/rspec1/latest/rng.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:468d3287781b465dcdb6fa02563ee9522fc7cde8a0d753d7862d7f591add6874
|
| 3 |
+
size 10501
|
runs/rspec1/latest/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52505970ce43f53a04aaadc70899b3b28a93790f8dcd6722810fd182d18a9d25
|
| 3 |
+
size 1401
|
runs/rspec1/latest/twin_config.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "Qwen/Qwen2.5-Math-1.5B",
|
| 3 |
+
"bidirectional": false,
|
| 4 |
+
"teacher_style": "solution_after_raw",
|
| 5 |
+
"lora_rank": 32,
|
| 6 |
+
"lora_alpha": 32.0,
|
| 7 |
+
"step": 90,
|
| 8 |
+
"val_metrics": {
|
| 9 |
+
"ce_encoder": 0.13284284267441218,
|
| 10 |
+
"ce_natural": 0.17520572880456878,
|
| 11 |
+
"ce_delta": -0.04236288613015662,
|
| 12 |
+
"ce_delta_easy": -0.04727620755484103,
|
| 13 |
+
"ce_delta_medium": -0.04218123273466182,
|
| 14 |
+
"ce_delta_very_hard": -0.02838024196957731,
|
| 15 |
+
"ce_delta_hard": -0.03736854335131856
|
| 16 |
+
}
|
| 17 |
+
}
|