s2_hard_06f: checkpoints exit
Browse files- runs/s2_hard_06f/best/lora_state.pt +3 -0
- runs/s2_hard_06f/best/twin_config.json +17 -0
- runs/s2_hard_06f/latest/lora_state.pt +3 -0
- runs/s2_hard_06f/latest/optimizer.pt +3 -0
- runs/s2_hard_06f/latest/rng.pt +3 -0
- runs/s2_hard_06f/latest/scheduler.pt +3 -0
- runs/s2_hard_06f/latest/twin_config.json +17 -0
runs/s2_hard_06f/best/lora_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7d21909afd2cdcb74628f76d225548a8db64da1574c7b3e6d433efdf6776cf7
|
| 3 |
+
size 147843609
|
runs/s2_hard_06f/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": 20,
|
| 8 |
+
"val_metrics": {
|
| 9 |
+
"ce_encoder": 0.1420903696565319,
|
| 10 |
+
"ce_natural": 0.17520572880456878,
|
| 11 |
+
"ce_delta": -0.033115359148036885,
|
| 12 |
+
"ce_delta_easy": -0.03664232930973606,
|
| 13 |
+
"ce_delta_medium": -0.03301396720329727,
|
| 14 |
+
"ce_delta_very_hard": -0.025415260304161004,
|
| 15 |
+
"ce_delta_hard": -0.028261169333241354
|
| 16 |
+
}
|
| 17 |
+
}
|
runs/s2_hard_06f/latest/lora_state.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:326a4edf632e0effee98ea38624c356f42032f3ebff708f81d4abe85ea47b87a
|
| 3 |
+
size 147843609
|
runs/s2_hard_06f/latest/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c4ec8a46c6bedea18130443336c2ecb30d0e10145a31f0d8de8c1b047fe2604
|
| 3 |
+
size 295766139
|
runs/s2_hard_06f/latest/rng.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a45d42e726a293c2ab9e4acea5e880ac5f4fa0094642f840be0f4c2fca24c7fb
|
| 3 |
+
size 10565
|
runs/s2_hard_06f/latest/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:47a9278d7319e08a805c476f641dddf3161820c6c90f5806c98d6d8a83b5ce90
|
| 3 |
+
size 1401
|
runs/s2_hard_06f/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": 40,
|
| 8 |
+
"val_metrics": {
|
| 9 |
+
"ce_encoder": 0.14365369488790744,
|
| 10 |
+
"ce_natural": 0.17520572880456878,
|
| 11 |
+
"ce_delta": -0.03155203391666134,
|
| 12 |
+
"ce_delta_easy": -0.03426331211314217,
|
| 13 |
+
"ce_delta_medium": -0.03222696882035022,
|
| 14 |
+
"ce_delta_very_hard": -0.022854911883835087,
|
| 15 |
+
"ce_delta_hard": -0.027748809194592407
|
| 16 |
+
}
|
| 17 |
+
}
|