Instructions to use LEAP-LAB-KUS/jevkt-checkpoints with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use LEAP-LAB-KUS/jevkt-checkpoints with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
meta_only podA e16 (configs + train logs; weights not backed up)
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/calibration.json +29 -0
- meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/judge_config.json +172 -0
- meta_only/podA/e16/S1-Dmix-oth/train_log.jsonl +15 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/calibration.json +29 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/judge_config.json +164 -0
- meta_only/podA/e16/S1-E19-a1-n64k-syn/train_log.jsonl +222 -0
- meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/calibration.json +29 -0
- meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/judge_config.json +164 -0
- meta_only/podA/e16/S1-F2syn4-rp/train_log.jsonl +15 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/calibration.json +29 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/judge_config.json +164 -0
- meta_only/podA/e16/S1-G-d4096-n16k-syn/train_log.jsonl +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/calibration.json +29 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/judge_config.json +164 -0
- meta_only/podA/e16/S1-G-d512-n16k-a025-syn/train_log.jsonl +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/calibration.json +29 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/judge_config.json +164 -0
- meta_only/podA/e16/S1-G-d512-n16k-a075-syn/train_log.jsonl +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/calibration.json +29 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/judge_config.json +164 -0
- meta_only/podA/e16/S1-G-d512-n16k-syn/train_log.jsonl +57 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json +99 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/calibration.json +29 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/judge_config.json +164 -0
- meta_only/podA/e16/S1-G-d64-n16k-syn/train_log.jsonl +57 -0
- meta_only/podA/e16/S1-O1-near/ckpt-4000/adapter/adapter_config.json +57 -0
- meta_only/podA/e16/S1-O1-near/ckpt-4000/adapter_vllm/adapter_config.json +58 -0
meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"in_proj_z",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"up_proj",
|
| 41 |
+
"down_proj",
|
| 42 |
+
"v_proj",
|
| 43 |
+
"in_proj_qkv",
|
| 44 |
+
"out_proj",
|
| 45 |
+
"q_proj",
|
| 46 |
+
"o_proj",
|
| 47 |
+
"k_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.00023752229753881693,
|
| 4 |
+
-0.00023777234309818596,
|
| 5 |
+
-0.0026873855385929346,
|
| 6 |
+
-0.0009875429095700383,
|
| 7 |
+
0.000309181836200878,
|
| 8 |
+
0.0018904521130025387,
|
| 9 |
+
0.0012668004492297769,
|
| 10 |
+
-0.002606811700388789,
|
| 11 |
+
-0.0005216100835241377,
|
| 12 |
+
-8.852832252159715e-05,
|
| 13 |
+
7.695446402067319e-05,
|
| 14 |
+
0.0005087539320811629,
|
| 15 |
+
0.00013379324809648097,
|
| 16 |
+
-0.0016750461654737592,
|
| 17 |
+
-0.0031597313936799765,
|
| 18 |
+
-0.005082675721496344,
|
| 19 |
+
-0.005936882458627224,
|
| 20 |
+
-0.004397138021886349,
|
| 21 |
+
-0.004939138423651457,
|
| 22 |
+
-0.0049933986738324165,
|
| 23 |
+
-0.004232514649629593,
|
| 24 |
+
-0.003407310228794813,
|
| 25 |
+
-0.005612242501229048,
|
| 26 |
+
-0.003486078465357423
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 1.0794936990788948,
|
| 5 |
+
"choice": 1.0057343784259787,
|
| 6 |
+
"score": 0.9117796091260059
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 665,
|
| 11 |
+
"T": 1.0794936990788948,
|
| 12 |
+
"ce_before": 0.4433257281780243,
|
| 13 |
+
"ce_after": 0.44257810711860657
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 371,
|
| 17 |
+
"T": 1.0057343784259787,
|
| 18 |
+
"ce_before": 0.5199505090713501,
|
| 19 |
+
"ce_after": 0.5199446678161621
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 164,
|
| 23 |
+
"T": 0.9117796091260059,
|
| 24 |
+
"ce_before": 1.1686280965805054,
|
| 25 |
+
"ce_after": 1.1667523384094238
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-Dmix-oth/ckpt-4000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-Dmix-oth/ckpt-4000/judge_config.json
ADDED
|
@@ -0,0 +1,172 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-Dmix-oth/ckpt-4000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e17/D-mix",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "D-mix",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "fdd4417db92ae0992db99cb0dee67868b8f2b35eecc407eeb63ffbc77c5ec1fd",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 125 steps",
|
| 128 |
+
"total_rows": 4000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-4000",
|
| 137 |
+
"rows_seen": 4000,
|
| 138 |
+
"replay_seen": 210,
|
| 139 |
+
"step": 125,
|
| 140 |
+
"tokens": 1687007,
|
| 141 |
+
"tok_per_s": 773.0954626723776,
|
| 142 |
+
"rows_per_s": 1.9292936510012286,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.4433257281780243,
|
| 147 |
+
"noul_T": 1.0794936990788948,
|
| 148 |
+
"choice_ce": 0.5199505090713501,
|
| 149 |
+
"choice_T": 1.0057343784259787,
|
| 150 |
+
"score_ce": 1.1686280965805054,
|
| 151 |
+
"score_T": 0.9117796091260059,
|
| 152 |
+
"noul_auc": 0.8998100955648126,
|
| 153 |
+
"noul_acc": 0.8339100346020761,
|
| 154 |
+
"noul_f1": 0.7018633540372671,
|
| 155 |
+
"noul_nll": 0.3474350189375224,
|
| 156 |
+
"noul_ece": 0.06752587609946935,
|
| 157 |
+
"noul_brier": 0.10720614748448454,
|
| 158 |
+
"noul_sharp": 0.292166531085968,
|
| 159 |
+
"noul_auc_boolq": 0.9766550522648084,
|
| 160 |
+
"noul_auc_forecast": 0.805982905982906,
|
| 161 |
+
"noul_auc_kuairec": 0.6080729166666667,
|
| 162 |
+
"choice_top1": 0.8713826366559485,
|
| 163 |
+
"choice_nll": 0.49305495619773865,
|
| 164 |
+
"choice_brier": 0.0415298193693161,
|
| 165 |
+
"choice_ece": 0.1062927239385832,
|
| 166 |
+
"score_rps": 0.01090323831886053,
|
| 167 |
+
"entropy": 0.5342580676078796
|
| 168 |
+
},
|
| 169 |
+
"dropped_long": 0,
|
| 170 |
+
"peak_gib": 18.235705852508545
|
| 171 |
+
}
|
| 172 |
+
}
|
meta_only/podA/e16/S1-Dmix-oth/train_log.jsonl
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "17:46:05", "event": "start", "data": "data/e17/D-mix", "data_manifest": {"arm": "D-mix", "name": null, "train_sha256": "fdd4417db92ae0992db99cb0dee67868b8f2b35eecc407eeb63ffbc77c5ec1fd", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 125 steps", "total_rows": 4000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 118.53579592704773}
|
| 2 |
+
{"time": "17:48:48", "event": "step", "step": 10, "rows_seen": 320, "lr": 9.923040269974204e-05, "tok_per_s": 822.9121269084353, "loss": 0.6263644024729729, "entropy": 0.5704365596175194, "sharp": 0.23654969428714953}
|
| 3 |
+
{"time": "17:51:27", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.55208574963224e-05, "tok_per_s": 838.8521427816331, "loss": 0.560571663081646, "entropy": 0.5510556921362877, "sharp": 0.28402155786752703}
|
| 4 |
+
{"time": "17:54:08", "event": "step", "step": 30, "rows_seen": 960, "lr": 8.897432118673851e-05, "tok_per_s": 837.7017477751401, "loss": 0.5879819840192795, "entropy": 0.5465441420674324, "sharp": 0.3122738411552028}
|
| 5 |
+
{"time": "17:56:44", "event": "step", "step": 40, "rows_seen": 1280, "lr": 8.002250216471678e-05, "tok_per_s": 839.364263087006, "loss": 0.5923318117856979, "entropy": 0.5699746757745743, "sharp": 0.2758946667114894}
|
| 6 |
+
{"time": "17:59:25", "event": "step", "step": 50, "rows_seen": 1600, "lr": 6.925572411108201e-05, "tok_per_s": 842.149273405047, "loss": 0.5427005141973495, "entropy": 0.5602971985936165, "sharp": 0.27341020420977946}
|
| 7 |
+
{"time": "18:02:03", "event": "step", "step": 60, "rows_seen": 1920, "lr": 5.738399736402462e-05, "tok_per_s": 842.8455921641022, "loss": 0.49934471398591995, "entropy": 0.45104507356882095, "sharp": 0.30936462628213984}
|
| 8 |
+
{"time": "18:04:45", "event": "step", "step": 70, "rows_seen": 2240, "lr": 4.519019760655705e-05, "tok_per_s": 844.3519974482363, "loss": 0.5480375811457634, "entropy": 0.5791695453226566, "sharp": 0.2608003098713724}
|
| 9 |
+
{"time": "18:07:23", "event": "step", "step": 80, "rows_seen": 2560, "lr": 3.3478439481564835e-05, "tok_per_s": 843.9358426214778, "loss": 0.5300071761012077, "entropy": 0.5281031131744385, "sharp": 0.25628655835201863}
|
| 10 |
+
{"time": "18:10:05", "event": "step", "step": 90, "rows_seen": 2880, "lr": 2.3021049613197416e-05, "tok_per_s": 845.2872436715257, "loss": 0.6388718858361244, "entropy": 0.593481108546257, "sharp": 0.2737026842016923}
|
| 11 |
+
{"time": "18:12:45", "event": "step", "step": 100, "rows_seen": 3200, "lr": 1.4507635874863332e-05, "tok_per_s": 845.188219371946, "loss": 0.5265334323048592, "entropy": 0.5574656650424004, "sharp": 0.2961363666935971}
|
| 12 |
+
{"time": "18:13:44", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.45032545924186707, "noul_T": 1.0676764845192572, "choice_ce": 0.48728087544441223, "choice_T": 0.969992721858492, "score_ce": 1.208053708076477, "score_T": 0.9679279597880692, "noul_auc": 0.891699604743083, "noul_acc": 0.8442906574394463, "noul_f1": 0.6896551724137931, "noul_nll": 0.3525707626741083, "noul_ece": 0.08735265002737407, "noul_brier": 0.10730570425762773, "noul_sharp": 0.28910863399505615, "noul_auc_boolq": 0.971774193548387, "noul_auc_forecast": 0.8325961963732862, "noul_auc_kuairec": 0.5714285714285714, "choice_top1": 0.8910256410256411, "choice_nll": 0.4872809052467346, "choice_brier": 0.04486461728811264, "choice_ece": 0.11063725768755643, "score_rps": 0.010897948406636715, "entropy": 0.5398765802383423}
|
| 13 |
+
{"time": "18:16:24", "event": "step", "step": 110, "rows_seen": 3520, "lr": 8.49961150805399e-06, "tok_per_s": 818.0331634808311, "loss": 0.5102563574910164, "entropy": 0.5306698828935623, "sharp": 0.32584547055395024}
|
| 14 |
+
{"time": "18:19:00", "event": "step", "step": 120, "rows_seen": 3840, "lr": 5.3931729785351795e-06, "tok_per_s": 819.8992897306567, "loss": 0.5645226389169693, "entropy": 0.5643603205680847, "sharp": 0.30647113134986476}
|
| 15 |
+
{"time": "18:22:28", "event": "ckpt", "tag": "ckpt-4000", "rows_seen": 4000, "replay_seen": 210, "step": 125, "tokens": 1687007, "tok_per_s": 773.0954626723776, "rows_per_s": 1.9292936510012286, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.4433257281780243, "noul_T": 1.0794936990788948, "choice_ce": 0.5199505090713501, "choice_T": 1.0057343784259787, "score_ce": 1.1686280965805054, "score_T": 0.9117796091260059, "noul_auc": 0.8998100955648126, "noul_acc": 0.8339100346020761, "noul_f1": 0.7018633540372671, "noul_nll": 0.3474350189375224, "noul_ece": 0.06752587609946935, "noul_brier": 0.10720614748448454, "noul_sharp": 0.292166531085968, "noul_auc_boolq": 0.9766550522648084, "noul_auc_forecast": 0.805982905982906, "noul_auc_kuairec": 0.6080729166666667, "choice_top1": 0.8713826366559485, "choice_nll": 0.49305495619773865, "choice_brier": 0.0415298193693161, "choice_ece": 0.1062927239385832, "score_rps": 0.01090323831886053, "entropy": 0.5342580676078796}, "dropped_long": 0, "peak_gib": 18.235705852508545}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"up_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"q_proj",
|
| 42 |
+
"gate_proj",
|
| 43 |
+
"out_proj",
|
| 44 |
+
"v_proj",
|
| 45 |
+
"in_proj_z",
|
| 46 |
+
"o_proj",
|
| 47 |
+
"in_proj_qkv"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0008289185934700072,
|
| 4 |
+
-0.0008292109705507755,
|
| 5 |
+
-0.0032280124723911285,
|
| 6 |
+
-0.0023483969271183014,
|
| 7 |
+
-0.0009108883095905185,
|
| 8 |
+
0.0018274933099746704,
|
| 9 |
+
0.0009582695201970637,
|
| 10 |
+
-0.0032973592169582844,
|
| 11 |
+
-0.000525892071891576,
|
| 12 |
+
-0.00012704676191788167,
|
| 13 |
+
8.483487908961251e-05,
|
| 14 |
+
0.0007388858357444406,
|
| 15 |
+
-0.0010332216043025255,
|
| 16 |
+
-0.00689094141125679,
|
| 17 |
+
-0.007822331972420216,
|
| 18 |
+
-0.006899011787027121,
|
| 19 |
+
-0.0065477751195430756,
|
| 20 |
+
-0.005011466331779957,
|
| 21 |
+
-0.005292436107993126,
|
| 22 |
+
-0.005928185768425465,
|
| 23 |
+
-0.0032564448192715645,
|
| 24 |
+
-0.004396967589855194,
|
| 25 |
+
-0.0063032531179487705,
|
| 26 |
+
-0.0040069688111543655
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8028773328820736,
|
| 5 |
+
"choice": 0.9985050815668384,
|
| 6 |
+
"score": 1.0349116888300849
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8028773328820736,
|
| 12 |
+
"ce_before": 0.5864824652671814,
|
| 13 |
+
"ce_after": 0.5833194255828857
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 0.9985050815668384,
|
| 18 |
+
"ce_before": 0.6727158427238464,
|
| 19 |
+
"ce_after": 0.6727151870727539
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 1.0349116888300849,
|
| 24 |
+
"ce_before": 1.0665595531463623,
|
| 25 |
+
"ce_after": 1.0662150382995605
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-E19-a1-n64k-syn/ckpt-64000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/ckpt-64000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-E19-a1-n64k-syn/ckpt-64000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e19/E19-a1-n64k",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "E19-a1-n64k",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "8837a87e06306751893fe83ba4f2ce3a2f3ce60ceca61f9e7fef3823de33bf8b",
|
| 110 |
+
"n_train": 64000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 2000 steps",
|
| 128 |
+
"total_rows": 64000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-64000",
|
| 137 |
+
"rows_seen": 64000,
|
| 138 |
+
"replay_seen": 3368,
|
| 139 |
+
"step": 2000,
|
| 140 |
+
"tokens": 35733211,
|
| 141 |
+
"tok_per_s": 861.0871689455226,
|
| 142 |
+
"rows_per_s": 1.623411912087117,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5864824652671814,
|
| 147 |
+
"noul_T": 0.8028773328820736,
|
| 148 |
+
"choice_ce": 0.6727158427238464,
|
| 149 |
+
"choice_T": 0.9985050815668384,
|
| 150 |
+
"score_ce": 1.0665595531463623,
|
| 151 |
+
"score_T": 1.0349116888300849,
|
| 152 |
+
"noul_auc": 0.7217706666666668,
|
| 153 |
+
"noul_acc": 0.686,
|
| 154 |
+
"noul_f1": 0.751188589540412,
|
| 155 |
+
"noul_nll": 0.5977250373587012,
|
| 156 |
+
"noul_ece": 0.08486032821238039,
|
| 157 |
+
"noul_brier": 0.2051691787868441,
|
| 158 |
+
"noul_sharp": 0.1511927843093872,
|
| 159 |
+
"entropy": 0.6252002716064453
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.235705852508545
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-E19-a1-n64k-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,222 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "22:16:37", "event": "start", "data": "data/e19/E19-a1-n64k", "data_manifest": {"arm": "E19-a1-n64k", "name": null, "train_sha256": "8837a87e06306751893fe83ba4f2ce3a2f3ce60ceca61f9e7fef3823de33bf8b", "n_train": 64000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 2000 steps", "total_rows": 64000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 94.76827621459961}
|
| 2 |
+
{"time": "22:19:58", "event": "step", "step": 10, "rows_seen": 320, "lr": 1.8333333333333333e-05, "tok_per_s": 879.8281962821995, "loss": 0.6295046769082546, "entropy": 0.6324522346258163, "sharp": 0.13728257454931736}
|
| 3 |
+
{"time": "22:23:19", "event": "step", "step": 20, "rows_seen": 640, "lr": 3.5e-05, "tok_per_s": 885.2188230557209, "loss": 0.644701637327671, "entropy": 0.6448138281702995, "sharp": 0.1410806979984045}
|
| 4 |
+
{"time": "22:26:28", "event": "step", "step": 30, "rows_seen": 960, "lr": 5.166666666666667e-05, "tok_per_s": 891.0816498264118, "loss": 0.6521275788545609, "entropy": 0.6206205263733864, "sharp": 0.15259388834238052}
|
| 5 |
+
{"time": "22:29:48", "event": "step", "step": 40, "rows_seen": 1280, "lr": 6.833333333333333e-05, "tok_per_s": 895.1496204492779, "loss": 0.6161444261670113, "entropy": 0.6178058609366417, "sharp": 0.16081522218883038}
|
| 6 |
+
{"time": "22:33:07", "event": "step", "step": 50, "rows_seen": 1600, "lr": 8.5e-05, "tok_per_s": 897.8612788529334, "loss": 0.7212060801684856, "entropy": 0.5909525919705629, "sharp": 0.18462728895246983}
|
| 7 |
+
{"time": "22:36:28", "event": "step", "step": 60, "rows_seen": 1920, "lr": 0.0001, "tok_per_s": 900.4602639406766, "loss": 0.5949314124882221, "entropy": 0.5915489420294762, "sharp": 0.20434223487973213}
|
| 8 |
+
{"time": "22:39:41", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.999377197411753e-05, "tok_per_s": 899.9785302273847, "loss": 0.5792611092329025, "entropy": 0.48742804303765297, "sharp": 0.24862276017665863}
|
| 9 |
+
{"time": "22:43:00", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.997508952966197e-05, "tok_per_s": 900.2097428800525, "loss": 0.5567303448915482, "entropy": 0.5903446711599827, "sharp": 0.20334683824330568}
|
| 10 |
+
{"time": "22:46:28", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.994395756578056e-05, "tok_per_s": 900.7512363602707, "loss": 0.6225418150424957, "entropy": 0.5749045498669147, "sharp": 0.20486978814005852}
|
| 11 |
+
{"time": "22:49:44", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.990038424629131e-05, "tok_per_s": 900.369465263043, "loss": 0.5931683257222176, "entropy": 0.6158639714121819, "sharp": 0.1678940560668707}
|
| 12 |
+
{"time": "22:51:07", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5941492319107056, "noul_T": 0.8441379369858887, "noul_auc": 0.7347555315522102, "noul_acc": 0.71, "noul_f1": 0.7702060221870047, "noul_nll": 0.5886883767992258, "noul_ece": 0.08399680578708647, "noul_brier": 0.20087797398754728, "noul_sharp": 0.16528591513633728, "entropy": 0.6168355345726013}
|
| 13 |
+
{"time": "22:54:28", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.984438099754206e-05, "tok_per_s": 867.4311964952135, "loss": 0.5975734703242779, "entropy": 0.5767129883170128, "sharp": 0.2061951719224453}
|
| 14 |
+
{"time": "22:57:47", "event": "step", "step": 120, "rows_seen": 3840, "lr": 9.977596250541417e-05, "tok_per_s": 869.9839570737339, "loss": 0.6067100986838341, "entropy": 0.561366856098175, "sharp": 0.21501660719513893}
|
| 15 |
+
{"time": "23:01:10", "event": "step", "step": 130, "rows_seen": 4160, "lr": 9.969514671147145e-05, "tok_per_s": 872.3897959559512, "loss": 0.5839385762810707, "entropy": 0.6382153034210205, "sharp": 0.12768653221428394}
|
| 16 |
+
{"time": "23:04:32", "event": "step", "step": 140, "rows_seen": 4480, "lr": 9.96019548082552e-05, "tok_per_s": 874.5012354375773, "loss": 0.5903449468314648, "entropy": 0.5751801058650017, "sharp": 0.20898077730089426}
|
| 17 |
+
{"time": "23:07:56", "event": "step", "step": 150, "rows_seen": 4800, "lr": 9.949641123372692e-05, "tok_per_s": 875.9995708882053, "loss": 0.6324847638607025, "entropy": 0.5247549340128899, "sharp": 0.2591211348772049}
|
| 18 |
+
{"time": "23:11:17", "event": "step", "step": 160, "rows_seen": 5120, "lr": 9.937854366485982e-05, "tok_per_s": 877.6046686209174, "loss": 0.6013332903385162, "entropy": 0.6032251119613647, "sharp": 0.18289333581924438}
|
| 19 |
+
{"time": "23:14:37", "event": "step", "step": 170, "rows_seen": 5440, "lr": 9.924838301038104e-05, "tok_per_s": 878.7638186454827, "loss": 0.6544026359915733, "entropy": 0.5716826356947422, "sharp": 0.19931085780262947}
|
| 20 |
+
{"time": "23:17:58", "event": "step", "step": 180, "rows_seen": 5760, "lr": 9.910596340266635e-05, "tok_per_s": 880.1927363993137, "loss": 0.5470241568982601, "entropy": 0.5960400179028511, "sharp": 0.19170998223125935}
|
| 21 |
+
{"time": "23:21:15", "event": "step", "step": 190, "rows_seen": 6080, "lr": 9.895132218878961e-05, "tok_per_s": 881.3103660518425, "loss": 0.5339637324213982, "entropy": 0.6053320877254009, "sharp": 0.1852043718099594}
|
| 22 |
+
{"time": "23:24:36", "event": "step", "step": 200, "rows_seen": 6400, "lr": 9.8784499920729e-05, "tok_per_s": 882.5939040029452, "loss": 0.6164066307246685, "entropy": 0.5684748440980911, "sharp": 0.22272171080112457}
|
| 23 |
+
{"time": "23:25:59", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.669293224811554, "noul_T": 1.2029442194229711, "noul_auc": 0.7173718587028067, "noul_acc": 0.598, "noul_f1": 0.592292089249493, "noul_nll": 0.6643123382627965, "noul_ece": 0.16651845568418502, "noul_brier": 0.23597738877164245, "noul_sharp": 0.10692768543958664, "entropy": 0.6583645343780518}
|
| 24 |
+
{"time": "23:29:11", "event": "step", "step": 210, "rows_seen": 6720, "lr": 9.860554034473306e-05, "tok_per_s": 866.5258959366943, "loss": 0.5758124673739076, "entropy": 0.5514099942520261, "sharp": 0.22141501866281033}
|
| 25 |
+
{"time": "23:32:29", "event": "step", "step": 220, "rows_seen": 7040, "lr": 9.841449038984903e-05, "tok_per_s": 868.0751848766342, "loss": 0.6564287673681974, "entropy": 0.6063540726900101, "sharp": 0.18049338273704052}
|
| 26 |
+
{"time": "23:35:49", "event": "step", "step": 230, "rows_seen": 7360, "lr": 9.821140015561645e-05, "tok_per_s": 869.4628662185197, "loss": 0.5857905689626932, "entropy": 0.5639754328876734, "sharp": 0.21251035295426846}
|
| 27 |
+
{"time": "23:39:07", "event": "step", "step": 240, "rows_seen": 7680, "lr": 9.79963228989294e-05, "tok_per_s": 870.506194831773, "loss": 0.5568301826715469, "entropy": 0.569310862571001, "sharp": 0.21442159824073315}
|
| 28 |
+
{"time": "23:42:31", "event": "step", "step": 250, "rows_seen": 8000, "lr": 9.776931502007089e-05, "tok_per_s": 871.0893530194276, "loss": 0.5874522179365158, "entropy": 0.5461196340620518, "sharp": 0.2357271034270525}
|
| 29 |
+
{"time": "23:45:46", "event": "step", "step": 260, "rows_seen": 8320, "lr": 9.753043604792285e-05, "tok_per_s": 872.0375493939796, "loss": 0.5829084496945143, "entropy": 0.5955034252256155, "sharp": 0.18169008195400238}
|
| 30 |
+
{"time": "23:48:59", "event": "step", "step": 270, "rows_seen": 8640, "lr": 9.727974862435562e-05, "tok_per_s": 872.9650850795064, "loss": 0.5483494400978088, "entropy": 0.5428918153047562, "sharp": 0.24139914475381374}
|
| 31 |
+
{"time": "23:52:25", "event": "step", "step": 280, "rows_seen": 8960, "lr": 9.701731848780134e-05, "tok_per_s": 874.3574056925535, "loss": 0.6192214414477348, "entropy": 0.6410646811127663, "sharp": 0.12074428889900446}
|
| 32 |
+
{"time": "23:55:37", "event": "step", "step": 290, "rows_seen": 9280, "lr": 9.674321445601502e-05, "tok_per_s": 875.4015101964078, "loss": 0.6048538275063038, "entropy": 0.5243420638144016, "sharp": 0.25408149138092995}
|
| 33 |
+
{"time": "23:58:57", "event": "step", "step": 300, "rows_seen": 9600, "lr": 9.645750840802845e-05, "tok_per_s": 876.1632358091671, "loss": 0.6349929682910442, "entropy": 0.6488283388316631, "sharp": 0.11365789268165827}
|
| 34 |
+
{"time": "00:00:19", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.6076900362968445, "noul_T": 0.5704677992597221, "noul_auc": 0.7356162765185399, "noul_acc": 0.708, "noul_f1": 0.7689873417721519, "noul_nll": 0.6069956250190734, "noul_ece": 0.11072315752506257, "noul_brier": 0.20850681405221444, "noul_sharp": 0.11592690646648407, "entropy": 0.6538520455360413}
|
| 35 |
+
{"time": "00:03:29", "event": "step", "step": 310, "rows_seen": 9920, "lr": 9.616027526530109e-05, "tok_per_s": 865.5685746410178, "loss": 0.5975884273648262, "entropy": 0.5664006546139717, "sharp": 0.21556021459400654}
|
| 36 |
+
{"time": "00:06:40", "event": "step", "step": 320, "rows_seen": 10240, "lr": 9.585159297207315e-05, "tok_per_s": 866.3836108234291, "loss": 0.675388753414154, "entropy": 0.5702600926160812, "sharp": 0.21910367906093597}
|
| 37 |
+
{"time": "00:09:57", "event": "step", "step": 330, "rows_seen": 10560, "lr": 9.553154247492623e-05, "tok_per_s": 867.5294545735793, "loss": 0.6008182633668184, "entropy": 0.6264041662216187, "sharp": 0.14019079133868217}
|
| 38 |
+
{"time": "00:13:21", "event": "step", "step": 340, "rows_seen": 10880, "lr": 9.520020770155628e-05, "tok_per_s": 868.5583302108193, "loss": 0.6273871958255768, "entropy": 0.5217002108693123, "sharp": 0.25571405701339245}
|
| 39 |
+
{"time": "00:16:32", "event": "step", "step": 350, "rows_seen": 11200, "lr": 9.485767553876519e-05, "tok_per_s": 869.5143762067335, "loss": 0.570355586707592, "entropy": 0.5637779533863068, "sharp": 0.2134623285382986}
|
| 40 |
+
{"time": "00:20:00", "event": "step", "step": 360, "rows_seen": 11520, "lr": 9.450403580967612e-05, "tok_per_s": 870.4135184646269, "loss": 0.5793403796851635, "entropy": 0.576744768768549, "sharp": 0.18689356371760368}
|
| 41 |
+
{"time": "00:23:21", "event": "step", "step": 370, "rows_seen": 11840, "lr": 9.413938125017902e-05, "tok_per_s": 871.2780550041474, "loss": 0.5755486413836479, "entropy": 0.5626277886331081, "sharp": 0.20399115234613419}
|
| 42 |
+
{"time": "00:26:31", "event": "step", "step": 380, "rows_seen": 12160, "lr": 9.376380748461219e-05, "tok_per_s": 871.9276514635242, "loss": 0.6146507300436497, "entropy": 0.5506707038730383, "sharp": 0.2205024715512991}
|
| 43 |
+
{"time": "00:29:47", "event": "step", "step": 390, "rows_seen": 12480, "lr": 9.337741300068645e-05, "tok_per_s": 872.6187219757546, "loss": 0.579678320325911, "entropy": 0.6338309049606323, "sharp": 0.14583980571478605}
|
| 44 |
+
{"time": "00:33:04", "event": "step", "step": 400, "rows_seen": 12800, "lr": 9.298029912365855e-05, "tok_per_s": 873.3371365414336, "loss": 0.5566546488553286, "entropy": 0.5875056274235249, "sharp": 0.20008304342627525}
|
| 45 |
+
{"time": "00:34:26", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.619740903377533, "noul_T": 0.5332617433059381, "noul_auc": 0.747835479569965, "noul_acc": 0.704, "noul_f1": 0.7581699346405228, "noul_nll": 0.6157594657242298, "noul_ece": 0.14022405427694323, "noul_brier": 0.2123595969865579, "noul_sharp": 0.1030149832367897, "entropy": 0.6624778509140015}
|
| 46 |
+
{"time": "00:37:44", "event": "step", "step": 410, "rows_seen": 13120, "lr": 9.257256998976022e-05, "tok_per_s": 865.4753381124939, "loss": 0.6311286725103855, "entropy": 0.539732102304697, "sharp": 0.23184069246053696}
|
| 47 |
+
{"time": "00:41:05", "event": "step", "step": 420, "rows_seen": 13440, "lr": 9.215433251889039e-05, "tok_per_s": 866.3084632522347, "loss": 0.6692369226366282, "entropy": 0.4708427060395479, "sharp": 0.2785058841109276}
|
| 48 |
+
{"time": "00:44:24", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.172569638657718e-05, "tok_per_s": 867.2465103566561, "loss": 0.596193952485919, "entropy": 0.6132407411932945, "sharp": 0.16161784529685974}
|
| 49 |
+
{"time": "00:47:44", "event": "step", "step": 440, "rows_seen": 14080, "lr": 9.128677399521752e-05, "tok_per_s": 867.6829595503679, "loss": 0.6028909310698509, "entropy": 0.5887715071439743, "sharp": 0.20368981081992388}
|
| 50 |
+
{"time": "00:51:06", "event": "step", "step": 450, "rows_seen": 14400, "lr": 9.083768044460145e-05, "tok_per_s": 868.6332466165472, "loss": 0.5378617122769356, "entropy": 0.5088745020329952, "sharp": 0.27278335578739643}
|
| 51 |
+
{"time": "00:54:21", "event": "step", "step": 460, "rows_seen": 14720, "lr": 9.037853350172924e-05, "tok_per_s": 869.1488504323095, "loss": 0.5857997015118599, "entropy": 0.6274540834128857, "sharp": 0.15940093249082565}
|
| 52 |
+
{"time": "00:57:34", "event": "step", "step": 470, "rows_seen": 15040, "lr": 8.990945356992909e-05, "tok_per_s": 869.8143725311548, "loss": 0.7208652403205633, "entropy": 0.5580282919108868, "sharp": 0.22304022498428822}
|
| 53 |
+
{"time": "01:00:54", "event": "step", "step": 480, "rows_seen": 15360, "lr": 8.943056365728335e-05, "tok_per_s": 870.4849895279173, "loss": 0.5892188530415297, "entropy": 0.5636716857552528, "sharp": 0.19815975986421108}
|
| 54 |
+
{"time": "01:04:14", "event": "step", "step": 490, "rows_seen": 15680, "lr": 8.894198934437185e-05, "tok_per_s": 871.3394462602696, "loss": 0.4664166606962681, "entropy": 0.5499792471528053, "sharp": 0.23838773369789124}
|
| 55 |
+
{"time": "01:07:35", "event": "step", "step": 500, "rows_seen": 16000, "lr": 8.844385875134053e-05, "tok_per_s": 871.8765799455861, "loss": 0.5988381206989288, "entropy": 0.6273887678980827, "sharp": 0.1437857523560524}
|
| 56 |
+
{"time": "01:08:58", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.6142338514328003, "noul_T": 0.5222142614554969, "noul_auc": 0.732780881335336, "noul_acc": 0.71, "noul_f1": 0.7709320695102686, "noul_nll": 0.6135297712683677, "noul_ece": 0.1171648921966553, "noul_brier": 0.21138333097635917, "noul_sharp": 0.1058560460805893, "entropy": 0.6607292890548706}
|
| 57 |
+
{"time": "01:12:16", "event": "step", "step": 510, "rows_seen": 16320, "lr": 8.793630250430419e-05, "tok_per_s": 865.6071765734309, "loss": 0.5168402716517448, "entropy": 0.5670865327119827, "sharp": 0.20703217759728432}
|
| 58 |
+
{"time": "01:15:35", "event": "step", "step": 520, "rows_seen": 16640, "lr": 8.741945370109202e-05, "tok_per_s": 866.4426664550654, "loss": 0.5569516010582447, "entropy": 0.5347343645989895, "sharp": 0.22369972430169582}
|
| 59 |
+
{"time": "01:18:51", "event": "step", "step": 530, "rows_seen": 16960, "lr": 8.689344787634503e-05, "tok_per_s": 866.8978365899027, "loss": 0.523026236332953, "entropy": 0.5534832812845707, "sharp": 0.22822879068553448}
|
| 60 |
+
{"time": "01:22:04", "event": "step", "step": 540, "rows_seen": 17280, "lr": 8.63584229659744e-05, "tok_per_s": 867.4947090836608, "loss": 0.5776505405083299, "entropy": 0.5941237211227417, "sharp": 0.19663067255169153}
|
| 61 |
+
{"time": "01:25:21", "event": "step", "step": 550, "rows_seen": 17600, "lr": 8.581451927099022e-05, "tok_per_s": 868.0582891046311, "loss": 0.6159417182207108, "entropy": 0.5471399649977684, "sharp": 0.22552569583058357}
|
| 62 |
+
{"time": "01:28:37", "event": "step", "step": 560, "rows_seen": 17920, "lr": 8.526187942070997e-05, "tok_per_s": 868.6809223115172, "loss": 0.5426895692944527, "entropy": 0.6184079684317112, "sharp": 0.16044112294912338}
|
| 63 |
+
{"time": "01:32:00", "event": "step", "step": 570, "rows_seen": 18240, "lr": 8.470064833535642e-05, "tok_per_s": 869.3292562290505, "loss": 0.5903506502509117, "entropy": 0.5530513413250446, "sharp": 0.2359584141522646}
|
| 64 |
+
{"time": "01:35:11", "event": "step", "step": 580, "rows_seen": 18560, "lr": 8.413097318805486e-05, "tok_per_s": 869.7762991895876, "loss": 0.5935558639466763, "entropy": 0.5706107541918755, "sharp": 0.2021002173423767}
|
| 65 |
+
{"time": "01:38:34", "event": "step", "step": 590, "rows_seen": 18880, "lr": 8.35530033662395e-05, "tok_per_s": 870.40233538899, "loss": 0.6606946848332882, "entropy": 0.60708799213171, "sharp": 0.1694428976625204}
|
| 66 |
+
{"time": "01:41:54", "event": "step", "step": 600, "rows_seen": 19200, "lr": 8.296689043247915e-05, "tok_per_s": 870.7949280420801, "loss": 0.6182748340070248, "entropy": 0.5225614011287689, "sharp": 0.2532625272870064}
|
| 67 |
+
{"time": "01:43:17", "event": "val", "step": 600, "rows_seen": 19200, "n": 500, "dropped_long": 0, "noul_ce": 0.6303961873054504, "noul_T": 0.8797516473175114, "noul_auc": 0.7319538910735683, "noul_acc": 0.67, "noul_f1": 0.7150259067357513, "noul_nll": 0.621625142082572, "noul_ece": 0.1316122469902039, "noul_brier": 0.21501984026742604, "noul_sharp": 0.12968496978282928, "entropy": 0.6418253183364868}
|
| 68 |
+
{"time": "01:46:39", "event": "step", "step": 610, "rows_seen": 19520, "lr": 8.237278808473261e-05, "tok_per_s": 865.446587173677, "loss": 0.5619433633983135, "entropy": 0.530576791614294, "sharp": 0.25496701896190643}
|
| 69 |
+
{"time": "01:49:55", "event": "step", "step": 620, "rows_seen": 19840, "lr": 8.177085211604405e-05, "tok_per_s": 866.0129471065833, "loss": 0.5948339104652405, "entropy": 0.6126631498336792, "sharp": 0.15992842242121696}
|
| 70 |
+
{"time": "01:53:13", "event": "step", "step": 630, "rows_seen": 20160, "lr": 8.116124037368893e-05, "tok_per_s": 866.5187120037444, "loss": 0.631188839673996, "entropy": 0.48113343119621277, "sharp": 0.27435916662216187}
|
| 71 |
+
{"time": "01:56:30", "event": "step", "step": 640, "rows_seen": 20480, "lr": 8.054411271778132e-05, "tok_per_s": 866.9587308888975, "loss": 0.6048442721366882, "entropy": 0.6388514116406441, "sharp": 0.14063527062535286}
|
| 72 |
+
{"time": "01:59:47", "event": "step", "step": 650, "rows_seen": 20800, "lr": 7.991963097935343e-05, "tok_per_s": 867.6484701182017, "loss": 0.620092511177063, "entropy": 0.5807441771030426, "sharp": 0.20726383104920387}
|
| 73 |
+
{"time": "02:03:04", "event": "step", "step": 660, "rows_seen": 21120, "lr": 7.928795891791805e-05, "tok_per_s": 868.2065853436262, "loss": 0.6018486227840185, "entropy": 0.5532626546919346, "sharp": 0.22126339748501778}
|
| 74 |
+
{"time": "02:06:27", "event": "step", "step": 670, "rows_seen": 21440, "lr": 7.864926217852567e-05, "tok_per_s": 868.71163155348, "loss": 0.6083822026848793, "entropy": 0.5470050293952227, "sharp": 0.2334184069186449}
|
| 75 |
+
{"time": "02:09:47", "event": "step", "step": 680, "rows_seen": 21760, "lr": 7.800370824832674e-05, "tok_per_s": 869.2162718107304, "loss": 0.6780160591006279, "entropy": 0.5959932506084442, "sharp": 0.18740460649132729}
|
| 76 |
+
{"time": "02:13:06", "event": "step", "step": 690, "rows_seen": 22080, "lr": 7.735146641265118e-05, "tok_per_s": 869.6178509220267, "loss": 0.5739633738994598, "entropy": 0.5414974093437195, "sharp": 0.24544446542859077}
|
| 77 |
+
{"time": "02:16:19", "event": "step", "step": 700, "rows_seen": 22400, "lr": 7.669270771061626e-05, "tok_per_s": 870.0519057965676, "loss": 0.5882686730474234, "entropy": 0.6329325139522552, "sharp": 0.14591061975806952}
|
| 78 |
+
{"time": "02:17:42", "event": "val", "step": 700, "rows_seen": 22400, "n": 500, "dropped_long": 0, "noul_ce": 0.6105464100837708, "noul_T": 0.626210241510394, "noul_auc": 0.7327302492784932, "noul_acc": 0.722, "noul_f1": 0.7790143084260731, "noul_nll": 0.6069155186414719, "noul_ece": 0.10125351572036743, "noul_brier": 0.20852994179838136, "noul_sharp": 0.11956991255283356, "entropy": 0.6506409645080566}
|
| 79 |
+
{"time": "02:20:59", "event": "step", "step": 710, "rows_seen": 22720, "lr": 7.602760489027442e-05, "tok_per_s": 865.5055801826919, "loss": 0.550527879036963, "entropy": 0.5766395851969719, "sharp": 0.19756381586194038}
|
| 80 |
+
{"time": "02:24:17", "event": "step", "step": 720, "rows_seen": 23040, "lr": 7.535633236331324e-05, "tok_per_s": 865.9845126617766, "loss": 0.5951694026589394, "entropy": 0.6160982809960842, "sharp": 0.1543780704960227}
|
| 81 |
+
{"time": "02:27:30", "event": "step", "step": 730, "rows_seen": 23360, "lr": 7.467906615931893e-05, "tok_per_s": 866.3924404761601, "loss": 0.633855689316988, "entropy": 0.6219516098499298, "sharp": 0.15601705200970173}
|
| 82 |
+
{"time": "02:30:49", "event": "step", "step": 740, "rows_seen": 23680, "lr": 7.399598387961561e-05, "tok_per_s": 866.8916241599842, "loss": 0.578365508466959, "entropy": 0.5408406406641006, "sharp": 0.23869755491614342}
|
| 83 |
+
{"time": "02:34:18", "event": "step", "step": 750, "rows_seen": 24000, "lr": 7.330726465069241e-05, "tok_per_s": 867.4893210422044, "loss": 0.6143578048795462, "entropy": 0.5767947286367416, "sharp": 0.2023455947637558}
|
| 84 |
+
{"time": "02:37:42", "event": "step", "step": 760, "rows_seen": 24320, "lr": 7.26130890772308e-05, "tok_per_s": 867.9118657805725, "loss": 0.5307466015219688, "entropy": 0.5833768546581268, "sharp": 0.1986555214971304}
|
| 85 |
+
{"time": "02:40:55", "event": "step", "step": 770, "rows_seen": 24640, "lr": 7.191363919474393e-05, "tok_per_s": 868.2437036355461, "loss": 0.5116982944309711, "entropy": 0.5428986474871635, "sharp": 0.24496513605117798}
|
| 86 |
+
{"time": "02:44:13", "event": "step", "step": 780, "rows_seen": 24960, "lr": 7.120909842184114e-05, "tok_per_s": 868.5859928680636, "loss": 0.6104172989726067, "entropy": 0.6192113421857357, "sharp": 0.1630136677995324}
|
| 87 |
+
{"time": "02:47:35", "event": "step", "step": 790, "rows_seen": 25280, "lr": 7.049965151212958e-05, "tok_per_s": 869.006270577969, "loss": 0.66473488509655, "entropy": 0.5529407486319542, "sharp": 0.233426284044981}
|
| 88 |
+
{"time": "02:50:54", "event": "step", "step": 800, "rows_seen": 25600, "lr": 6.978548450576568e-05, "tok_per_s": 869.3485126184794, "loss": 0.5488703474402428, "entropy": 0.5310305505990982, "sharp": 0.24531328491866589}
|
| 89 |
+
{"time": "02:52:17", "event": "val", "step": 800, "rows_seen": 25600, "n": 500, "dropped_long": 0, "noul_ce": 0.638276994228363, "noul_T": 1.0176164500746923, "noul_auc": 0.7272282324348958, "noul_acc": 0.682, "noul_f1": 0.7234782608695652, "noul_nll": 0.6276239794045687, "noul_ece": 0.13139118874073027, "noul_brier": 0.2178588700686464, "noul_sharp": 0.13625478744506836, "entropy": 0.6365290284156799}
|
| 90 |
+
{"time": "02:55:34", "event": "step", "step": 810, "rows_seen": 25920, "lr": 6.90667846806694e-05, "tok_per_s": 865.6229363497725, "loss": 0.6041722856462002, "entropy": 0.5914291515946388, "sharp": 0.19501657970249653}
|
| 91 |
+
{"time": "02:58:56", "event": "step", "step": 820, "rows_seen": 26240, "lr": 6.834374050341399e-05, "tok_per_s": 866.0530577880772, "loss": 0.563560102134943, "entropy": 0.5666959881782532, "sharp": 0.21063528768718243}
|
| 92 |
+
{"time": "03:02:18", "event": "step", "step": 830, "rows_seen": 26560, "lr": 6.76165415798036e-05, "tok_per_s": 866.559483225472, "loss": 0.5377663336694241, "entropy": 0.5200601629912853, "sharp": 0.2510240990668535}
|
| 93 |
+
{"time": "03:05:38", "event": "step", "step": 840, "rows_seen": 26880, "lr": 6.688537860515266e-05, "tok_per_s": 867.0208826963283, "loss": 0.6055592782795429, "entropy": 0.5769760720431805, "sharp": 0.21105707809329033}
|
| 94 |
+
{"time": "03:08:53", "event": "step", "step": 850, "rows_seen": 27200, "lr": 6.615044331427935e-05, "tok_per_s": 867.2944833508166, "loss": 0.5653587877750397, "entropy": 0.5743969902396202, "sharp": 0.22205298393964767}
|
| 95 |
+
{"time": "03:12:12", "event": "step", "step": 860, "rows_seen": 27520, "lr": 6.541192843122628e-05, "tok_per_s": 867.5986993692077, "loss": 0.631022222340107, "entropy": 0.6096626129001379, "sharp": 0.17655779980123043}
|
| 96 |
+
{"time": "03:15:33", "event": "step", "step": 870, "rows_seen": 27840, "lr": 6.467002761872219e-05, "tok_per_s": 867.9595345722736, "loss": 0.571856090798974, "entropy": 0.572668420150876, "sharp": 0.20693980436772108}
|
| 97 |
+
{"time": "03:18:52", "event": "step", "step": 880, "rows_seen": 28160, "lr": 6.392493542739706e-05, "tok_per_s": 868.3633923937507, "loss": 0.6348532177507877, "entropy": 0.6003972738981247, "sharp": 0.1778871100395918}
|
| 98 |
+
{"time": "03:22:12", "event": "step", "step": 890, "rows_seen": 28480, "lr": 6.317684724476465e-05, "tok_per_s": 868.7130416331082, "loss": 0.6752804294228554, "entropy": 0.5834106281399727, "sharp": 0.20822030492126942}
|
| 99 |
+
{"time": "03:25:29", "event": "step", "step": 900, "rows_seen": 28800, "lr": 6.242595924398558e-05, "tok_per_s": 869.0209644718431, "loss": 0.5100527014583349, "entropy": 0.5667705424129963, "sharp": 0.21761420741677284}
|
| 100 |
+
{"time": "03:26:52", "event": "val", "step": 900, "rows_seen": 28800, "n": 500, "dropped_long": 0, "noul_ce": 0.6089743971824646, "noul_T": 0.8338712716538229, "noul_auc": 0.7326796172216503, "noul_acc": 0.71, "noul_f1": 0.763458401305057, "noul_nll": 0.6014971361905337, "noul_ece": 0.09676947563886643, "noul_brier": 0.20592525957551377, "noul_sharp": 0.15064676105976105, "entropy": 0.6296687126159668}
|
| 101 |
+
{"time": "03:30:11", "event": "step", "step": 910, "rows_seen": 29120, "lr": 6.16724683324242e-05, "tok_per_s": 865.6109250096446, "loss": 0.5815437808632851, "entropy": 0.5851453710347414, "sharp": 0.19391375966370106}
|
| 102 |
+
{"time": "03:33:31", "event": "step", "step": 920, "rows_seen": 29440, "lr": 6.091657210001309e-05, "tok_per_s": 866.0296726475019, "loss": 0.5863425023853779, "entropy": 0.5812318921089172, "sharp": 0.202917474322021}
|
| 103 |
+
{"time": "03:36:47", "event": "step", "step": 930, "rows_seen": 29760, "lr": 6.015846876743859e-05, "tok_per_s": 866.4728045070311, "loss": 0.5620491616427898, "entropy": 0.5998508259654045, "sharp": 0.18881416786462069}
|
| 104 |
+
{"time": "03:40:07", "event": "step", "step": 940, "rows_seen": 30080, "lr": 5.9398357134160796e-05, "tok_per_s": 866.8402420579287, "loss": 0.5721940919756889, "entropy": 0.561129704117775, "sharp": 0.21118141897022724}
|
| 105 |
+
{"time": "03:43:23", "event": "step", "step": 950, "rows_seen": 30400, "lr": 5.8636436526281816e-05, "tok_per_s": 867.3039433650998, "loss": 0.5560663565993309, "entropy": 0.6124218106269836, "sharp": 0.16758470237255096}
|
| 106 |
+
{"time": "03:46:41", "event": "step", "step": 960, "rows_seen": 30720, "lr": 5.7872906744276015e-05, "tok_per_s": 867.6228007759487, "loss": 0.5580924153327942, "entropy": 0.4962543770670891, "sharp": 0.27149172872304916}
|
| 107 |
+
{"time": "03:49:58", "event": "step", "step": 970, "rows_seen": 31040, "lr": 5.7107968010595795e-05, "tok_per_s": 867.9916365902104, "loss": 0.6483019292354584, "entropy": 0.6160846725106239, "sharp": 0.1529483124613762}
|
| 108 |
+
{"time": "03:53:11", "event": "step", "step": 980, "rows_seen": 31360, "lr": 5.634182091716671e-05, "tok_per_s": 868.2999508179879, "loss": 0.5768284387886524, "entropy": 0.5517377890646458, "sharp": 0.22552032209932804}
|
| 109 |
+
{"time": "03:56:22", "event": "step", "step": 990, "rows_seen": 31680, "lr": 5.5574666372785714e-05, "tok_per_s": 868.543009992191, "loss": 0.5938916243612766, "entropy": 0.6289665922522545, "sharp": 0.15355420298874378}
|
| 110 |
+
{"time": "03:59:40", "event": "step", "step": 1000, "rows_seen": 32000, "lr": 5.48067055504363e-05, "tok_per_s": 868.8804100796154, "loss": 0.5699882619082928, "entropy": 0.5483336001634598, "sharp": 0.2411150448024273}
|
| 111 |
+
{"time": "04:01:02", "event": "val", "step": 1000, "rows_seen": 32000, "n": 500, "dropped_long": 0, "noul_ce": 0.6248049139976501, "noul_T": 0.8985312868871821, "noul_auc": 0.736206983848374, "noul_acc": 0.678, "noul_f1": 0.7170474516695958, "noul_nll": 0.6192462531328201, "noul_ece": 0.12381355214118958, "noul_brier": 0.2142520225590276, "noul_sharp": 0.14101095497608185, "entropy": 0.6357983350753784}
|
| 112 |
+
{"time": "04:04:19", "event": "step", "step": 1010, "rows_seen": 32320, "lr": 5.403813983453441e-05, "tok_per_s": 865.8019764295008, "loss": 0.5340769067406654, "entropy": 0.49972651433199644, "sharp": 0.2818912100046873}
|
| 113 |
+
{"time": "04:07:39", "event": "step", "step": 1020, "rows_seen": 32640, "lr": 5.3269170768118824e-05, "tok_per_s": 866.130998414858, "loss": 0.566925972700119, "entropy": 0.564256276935339, "sharp": 0.22859442606568336}
|
| 114 |
+
{"time": "04:10:55", "event": "step", "step": 1030, "rows_seen": 32960, "lr": 5.25e-05, "tok_per_s": 866.3906650846411, "loss": 0.5861629592254758, "entropy": 0.5962873306125402, "sharp": 0.1862189806997776}
|
| 115 |
+
{"time": "04:14:18", "event": "step", "step": 1040, "rows_seen": 33280, "lr": 5.173082923188119e-05, "tok_per_s": 866.833348641669, "loss": 0.5164126008749008, "entropy": 0.5392742082476616, "sharp": 0.2461160011589527}
|
| 116 |
+
{"time": "04:17:30", "event": "step", "step": 1050, "rows_seen": 33600, "lr": 5.0961860165465604e-05, "tok_per_s": 867.135943169979, "loss": 0.5485187470912933, "entropy": 0.5757041722536087, "sharp": 0.1988809872418642}
|
| 117 |
+
{"time": "04:20:47", "event": "step", "step": 1060, "rows_seen": 33920, "lr": 5.019329444956372e-05, "tok_per_s": 867.380690196777, "loss": 0.517763577401638, "entropy": 0.562677375972271, "sharp": 0.21104974299669266}
|
| 118 |
+
{"time": "04:24:08", "event": "step", "step": 1070, "rows_seen": 34240, "lr": 4.9425333627214304e-05, "tok_per_s": 867.7601555929893, "loss": 0.5897710435092449, "entropy": 0.5873810779303312, "sharp": 0.19086614064872265}
|
| 119 |
+
{"time": "04:27:28", "event": "step", "step": 1080, "rows_seen": 34560, "lr": 4.865817908283331e-05, "tok_per_s": 868.1166917841729, "loss": 0.5843389201909304, "entropy": 0.6066151857376099, "sharp": 0.1785845821723342}
|
| 120 |
+
{"time": "04:30:49", "event": "step", "step": 1090, "rows_seen": 34880, "lr": 4.7892031989404217e-05, "tok_per_s": 868.4530794585155, "loss": 0.6169258877635002, "entropy": 0.6287201717495918, "sharp": 0.14132500253617764}
|
| 121 |
+
{"time": "04:34:12", "event": "step", "step": 1100, "rows_seen": 35200, "lr": 4.7127093255723996e-05, "tok_per_s": 868.7922787121851, "loss": 0.5694097187370062, "entropy": 0.6099346205592155, "sharp": 0.17100903298705816}
|
| 122 |
+
{"time": "04:35:35", "event": "val", "step": 1100, "rows_seen": 35200, "n": 500, "dropped_long": 0, "noul_ce": 0.6074497699737549, "noul_T": 0.6890390195432704, "noul_auc": 0.7409157651347656, "noul_acc": 0.708, "noul_f1": 0.760655737704918, "noul_nll": 0.6031571755409241, "noul_ece": 0.10881973236799242, "noul_brier": 0.20683276337410592, "noul_sharp": 0.13114292919635773, "entropy": 0.6434270143508911}
|
| 123 |
+
{"time": "04:38:54", "event": "step", "step": 1110, "rows_seen": 35520, "lr": 4.636356347371821e-05, "tok_per_s": 865.9820024032896, "loss": 0.5474040796980262, "entropy": 0.5117474906146526, "sharp": 0.2584786154329777}
|
| 124 |
+
{"time": "04:42:11", "event": "step", "step": 1120, "rows_seen": 35840, "lr": 4.560164286583922e-05, "tok_per_s": 866.3355919045679, "loss": 0.560987826436758, "entropy": 0.5968900695443153, "sharp": 0.18466806039214134}
|
| 125 |
+
{"time": "04:45:30", "event": "step", "step": 1130, "rows_seen": 36160, "lr": 4.4841531232561406e-05, "tok_per_s": 866.6726485958843, "loss": 0.5249037835747004, "entropy": 0.577229768037796, "sharp": 0.21040616929531097}
|
| 126 |
+
{"time": "04:48:48", "event": "step", "step": 1140, "rows_seen": 36480, "lr": 4.4083427899986926e-05, "tok_per_s": 867.0238137484558, "loss": 0.6443790309131145, "entropy": 0.5617045871913433, "sharp": 0.22774842008948326}
|
| 127 |
+
{"time": "04:52:04", "event": "step", "step": 1150, "rows_seen": 36800, "lr": 4.3327531667575824e-05, "tok_per_s": 867.3781773528385, "loss": 0.6746242865920067, "entropy": 0.5511924587190151, "sharp": 0.2338515892624855}
|
| 128 |
+
{"time": "04:55:19", "event": "step", "step": 1160, "rows_seen": 37120, "lr": 4.257404075601443e-05, "tok_per_s": 867.6247154893954, "loss": 0.5841744216158986, "entropy": 0.5851995535194874, "sharp": 0.19160346314311028}
|
| 129 |
+
{"time": "04:58:34", "event": "step", "step": 1170, "rows_seen": 37440, "lr": 4.182315275523537e-05, "tok_per_s": 867.8479207041956, "loss": 0.5574506670236588, "entropy": 0.5442451201379299, "sharp": 0.2422637827694416}
|
| 130 |
+
{"time": "05:01:53", "event": "step", "step": 1180, "rows_seen": 37760, "lr": 4.1075064572602964e-05, "tok_per_s": 868.0730794520318, "loss": 0.5989802666008472, "entropy": 0.6013084352016449, "sharp": 0.18256116472184658}
|
| 131 |
+
{"time": "05:05:14", "event": "step", "step": 1190, "rows_seen": 38080, "lr": 4.0329972381277805e-05, "tok_per_s": 868.3802389068825, "loss": 0.5671998322941363, "entropy": 0.5493317758664489, "sharp": 0.23072188720107079}
|
| 132 |
+
{"time": "05:08:38", "event": "step", "step": 1200, "rows_seen": 38400, "lr": 3.958807156877371e-05, "tok_per_s": 868.6134227819826, "loss": 0.5607874467968941, "entropy": 0.5767529830336571, "sharp": 0.20110230892896652}
|
| 133 |
+
{"time": "05:10:00", "event": "val", "step": 1200, "rows_seen": 38400, "n": 500, "dropped_long": 0, "noul_ce": 0.5889299511909485, "noul_T": 0.8596232812950947, "noul_auc": 0.7376246814399756, "noul_acc": 0.722, "noul_f1": 0.7831513260530422, "noul_nll": 0.5829388448894024, "noul_ece": 0.056635709524154663, "noul_brier": 0.19802001940458538, "noul_sharp": 0.17348846793174744, "entropy": 0.6105198264122009}
|
| 134 |
+
{"time": "05:13:17", "event": "step", "step": 1210, "rows_seen": 38720, "lr": 3.884955668572066e-05, "tok_per_s": 866.0167387705078, "loss": 0.5979101583361626, "entropy": 0.5728268474340439, "sharp": 0.20886495523154736}
|
| 135 |
+
{"time": "05:16:34", "event": "step", "step": 1220, "rows_seen": 39040, "lr": 3.811462139484735e-05, "tok_per_s": 866.2515373707987, "loss": 0.5668299198150635, "entropy": 0.545787550508976, "sharp": 0.2324742078781128}
|
| 136 |
+
{"time": "05:19:50", "event": "step", "step": 1230, "rows_seen": 39360, "lr": 3.738345842019643e-05, "tok_per_s": 866.4838814880198, "loss": 0.5421960782259703, "entropy": 0.5743382880464196, "sharp": 0.20786747336387634}
|
| 137 |
+
{"time": "05:23:10", "event": "step", "step": 1240, "rows_seen": 39680, "lr": 3.665625949658603e-05, "tok_per_s": 866.7325489863255, "loss": 0.6211932189762592, "entropy": 0.5689419712871313, "sharp": 0.212791764177382}
|
| 138 |
+
{"time": "05:26:33", "event": "step", "step": 1250, "rows_seen": 40000, "lr": 3.5933215319330585e-05, "tok_per_s": 866.9305787264714, "loss": 0.5803083814680576, "entropy": 0.5402863323688507, "sharp": 0.2442592829465866}
|
| 139 |
+
{"time": "05:29:50", "event": "step", "step": 1260, "rows_seen": 40320, "lr": 3.521451549423435e-05, "tok_per_s": 867.1617179583308, "loss": 0.536202278919518, "entropy": 0.607521353289485, "sharp": 0.1830746103078127}
|
| 140 |
+
{"time": "05:33:08", "event": "step", "step": 1270, "rows_seen": 40640, "lr": 3.450034848787043e-05, "tok_per_s": 867.3822721903856, "loss": 0.5611704513430595, "entropy": 0.5578159168362617, "sharp": 0.219560869038105}
|
| 141 |
+
{"time": "05:36:30", "event": "step", "step": 1280, "rows_seen": 40960, "lr": 3.379090157815887e-05, "tok_per_s": 867.7121323594877, "loss": 0.6312098801136017, "entropy": 0.6213430240750313, "sharp": 0.15102044492959976}
|
| 142 |
+
{"time": "05:39:48", "event": "step", "step": 1290, "rows_seen": 41280, "lr": 3.3086360805256096e-05, "tok_per_s": 867.9605811131463, "loss": 0.623515959829092, "entropy": 0.5751582253724337, "sharp": 0.20888733305037022}
|
| 143 |
+
{"time": "05:43:15", "event": "step", "step": 1300, "rows_seen": 41600, "lr": 3.238691092276923e-05, "tok_per_s": 868.1946714939847, "loss": 0.570844542235136, "entropy": 0.5764611540362239, "sharp": 0.20984778180718422}
|
| 144 |
+
{"time": "05:44:38", "event": "val", "step": 1300, "rows_seen": 41600, "n": 500, "dropped_long": 0, "noul_ce": 0.6007070541381836, "noul_T": 0.8379441783335629, "noul_auc": 0.7372365023375134, "noul_acc": 0.702, "noul_f1": 0.7529021558872305, "noul_nll": 0.5950360209196806, "noul_ece": 0.0794775465130806, "noul_brier": 0.20355869228743173, "noul_sharp": 0.1577596217393875, "entropy": 0.6227014064788818}
|
| 145 |
+
{"time": "05:47:57", "event": "step", "step": 1310, "rows_seen": 41920, "lr": 3.169273534930759e-05, "tok_per_s": 865.7551569450357, "loss": 0.62377804890275, "entropy": 0.5168309845030308, "sharp": 0.26707932353019714}
|
| 146 |
+
{"time": "05:51:16", "event": "step", "step": 1320, "rows_seen": 42240, "lr": 3.10040161203844e-05, "tok_per_s": 866.0175437091998, "loss": 0.6280301474034786, "entropy": 0.5768657550215721, "sharp": 0.2072569550946355}
|
| 147 |
+
{"time": "05:54:28", "event": "step", "step": 1330, "rows_seen": 42560, "lr": 3.032093384068107e-05, "tok_per_s": 866.192853597795, "loss": 0.6169680468738079, "entropy": 0.575705099850893, "sharp": 0.20512978173792362}
|
| 148 |
+
{"time": "05:57:52", "event": "step", "step": 1340, "rows_seen": 42880, "lr": 2.9643667636686757e-05, "tok_per_s": 866.4723627294193, "loss": 0.5304111652076244, "entropy": 0.5600172095000744, "sharp": 0.2188215060159564}
|
| 149 |
+
{"time": "06:01:08", "event": "step", "step": 1350, "rows_seen": 43200, "lr": 2.8972395109725607e-05, "tok_per_s": 866.6554123017848, "loss": 0.535086190328002, "entropy": 0.5687444526702166, "sharp": 0.21896788105368614}
|
| 150 |
+
{"time": "06:04:28", "event": "step", "step": 1360, "rows_seen": 43520, "lr": 2.8307292289383776e-05, "tok_per_s": 866.8303305335215, "loss": 0.5771385058760643, "entropy": 0.5646099746227264, "sharp": 0.22288157045841217}
|
| 151 |
+
{"time": "06:07:50", "event": "step", "step": 1370, "rows_seen": 43840, "lr": 2.7648533587348822e-05, "tok_per_s": 867.1030438127983, "loss": 0.5699361003935337, "entropy": 0.5227292329072952, "sharp": 0.24444935098290443}
|
| 152 |
+
{"time": "06:11:15", "event": "step", "step": 1380, "rows_seen": 44160, "lr": 2.699629175167327e-05, "tok_per_s": 867.3477072918165, "loss": 0.580060139298439, "entropy": 0.5431000143289566, "sharp": 0.24059907719492912}
|
| 153 |
+
{"time": "06:14:39", "event": "step", "step": 1390, "rows_seen": 44480, "lr": 2.635073782147433e-05, "tok_per_s": 867.6050399687886, "loss": 0.5966040007770061, "entropy": 0.5633541010320187, "sharp": 0.21955593675374985}
|
| 154 |
+
{"time": "06:17:52", "event": "step", "step": 1400, "rows_seen": 44800, "lr": 2.571204108208194e-05, "tok_per_s": 867.7580814787839, "loss": 0.6264890441671014, "entropy": 0.6002929843962193, "sharp": 0.1784014394506812}
|
| 155 |
+
{"time": "06:19:15", "event": "val", "step": 1400, "rows_seen": 44800, "n": 500, "dropped_long": 0, "noul_ce": 0.6048054099082947, "noul_T": 0.7839859385207317, "noul_auc": 0.737439030564885, "noul_acc": 0.692, "noul_f1": 0.7407407407407407, "noul_nll": 0.5999438506662845, "noul_ece": 0.095223878800869, "noul_brier": 0.20569923290131492, "noul_sharp": 0.14640359580516815, "entropy": 0.6320462822914124}
|
| 156 |
+
{"time": "06:22:32", "event": "step", "step": 1410, "rows_seen": 45120, "lr": 2.5080369020646587e-05, "tok_per_s": 865.4147202486301, "loss": 0.5373427271842957, "entropy": 0.5949507281184196, "sharp": 0.19869450107216835}
|
| 157 |
+
{"time": "06:25:52", "event": "step", "step": 1420, "rows_seen": 45440, "lr": 2.4455887282218688e-05, "tok_per_s": 865.6004597637758, "loss": 0.6266276463866234, "entropy": 0.6115313768386841, "sharp": 0.17394527420401573}
|
| 158 |
+
{"time": "06:29:07", "event": "step", "step": 1430, "rows_seen": 45760, "lr": 2.383875962631108e-05, "tok_per_s": 865.8181087389024, "loss": 0.5567162670195103, "entropy": 0.5424675419926643, "sharp": 0.23107896372675896}
|
| 159 |
+
{"time": "06:32:23", "event": "step", "step": 1440, "rows_seen": 46080, "lr": 2.322914788395595e-05, "tok_per_s": 866.0650555008157, "loss": 0.6491935923695564, "entropy": 0.5880753248929977, "sharp": 0.1993061937391758}
|
| 160 |
+
{"time": "06:35:40", "event": "step", "step": 1450, "rows_seen": 46400, "lr": 2.2627211915267383e-05, "tok_per_s": 866.2261269502333, "loss": 0.5494246296584606, "entropy": 0.5415302254259586, "sharp": 0.24098924547433853}
|
| 161 |
+
{"time": "06:38:55", "event": "step", "step": 1460, "rows_seen": 46720, "lr": 2.2033109567520866e-05, "tok_per_s": 866.4105077658206, "loss": 0.586689904332161, "entropy": 0.5798527002334595, "sharp": 0.20355992764234543}
|
| 162 |
+
{"time": "06:42:12", "event": "step", "step": 1470, "rows_seen": 47040, "lr": 2.144699663376052e-05, "tok_per_s": 866.6264650658495, "loss": 0.5970914848148823, "entropy": 0.5527854263782501, "sharp": 0.2395586296916008}
|
| 163 |
+
{"time": "06:45:24", "event": "step", "step": 1480, "rows_seen": 47360, "lr": 2.0869026811945157e-05, "tok_per_s": 866.7739158599641, "loss": 0.5584869254380465, "entropy": 0.564186729490757, "sharp": 0.21074647642672062}
|
| 164 |
+
{"time": "06:48:45", "event": "step", "step": 1490, "rows_seen": 47680, "lr": 2.0299351664643596e-05, "tok_per_s": 867.0013723869188, "loss": 0.553637919947505, "entropy": 0.5442855469882488, "sharp": 0.24212833307683468}
|
| 165 |
+
{"time": "06:52:03", "event": "step", "step": 1500, "rows_seen": 48000, "lr": 1.973812057929005e-05, "tok_per_s": 867.2165525072193, "loss": 0.5874327160418034, "entropy": 0.5660212114453316, "sharp": 0.22206989210098982}
|
| 166 |
+
{"time": "06:53:25", "event": "val", "step": 1500, "rows_seen": 48000, "n": 500, "dropped_long": 0, "noul_ce": 0.6134161949157715, "noul_T": 0.9039978067471125, "noul_auc": 0.7354475029957299, "noul_acc": 0.68, "noul_f1": 0.7222222222222222, "noul_nll": 0.6078775519132614, "noul_ece": 0.10573660296201706, "noul_brier": 0.20934835315460265, "noul_sharp": 0.15412095189094543, "entropy": 0.6253897547721863}
|
| 167 |
+
{"time": "06:56:41", "event": "step", "step": 1510, "rows_seen": 48320, "lr": 1.9185480729009792e-05, "tok_per_s": 865.1184913141209, "loss": 0.5599006414413452, "entropy": 0.555024541914463, "sharp": 0.22217212244868279}
|
| 168 |
+
{"time": "06:59:57", "event": "step", "step": 1520, "rows_seen": 48640, "lr": 1.8641577034025627e-05, "tok_per_s": 865.3566902178209, "loss": 0.5344938226044178, "entropy": 0.6059355586767197, "sharp": 0.1761352838948369}
|
| 169 |
+
{"time": "07:03:18", "event": "step", "step": 1530, "rows_seen": 48960, "lr": 1.810655212365499e-05, "tok_per_s": 865.5749125396756, "loss": 0.5552831031382084, "entropy": 0.5546936765313148, "sharp": 0.21888944506645203}
|
| 170 |
+
{"time": "07:06:39", "event": "step", "step": 1540, "rows_seen": 49280, "lr": 1.7580546298907994e-05, "tok_per_s": 865.802099022887, "loss": 0.5533717945218086, "entropy": 0.5638622045516968, "sharp": 0.21822088584303856}
|
| 171 |
+
{"time": "07:10:02", "event": "step", "step": 1550, "rows_seen": 49600, "lr": 1.7063697495695813e-05, "tok_per_s": 866.0264452105991, "loss": 0.5595994740724564, "entropy": 0.5647246018052101, "sharp": 0.21260201930999756}
|
| 172 |
+
{"time": "07:13:25", "event": "step", "step": 1560, "rows_seen": 49920, "lr": 1.6556141248659472e-05, "tok_per_s": 866.2575624401504, "loss": 0.5422917529940605, "entropy": 0.5599235743284225, "sharp": 0.21861359290778637}
|
| 173 |
+
{"time": "07:16:39", "event": "step", "step": 1570, "rows_seen": 50240, "lr": 1.605801065562816e-05, "tok_per_s": 866.434708591743, "loss": 0.579589219763875, "entropy": 0.5927067399024963, "sharp": 0.1884201504290104}
|
| 174 |
+
{"time": "07:19:58", "event": "step", "step": 1580, "rows_seen": 50560, "lr": 1.556943634271667e-05, "tok_per_s": 866.6079575085386, "loss": 0.48137733340263367, "entropy": 0.5159203484654427, "sharp": 0.249346774071455}
|
| 175 |
+
{"time": "07:23:14", "event": "step", "step": 1590, "rows_seen": 50880, "lr": 1.5090546430070923e-05, "tok_per_s": 866.8001352622437, "loss": 0.5211900994181633, "entropy": 0.5357989966869354, "sharp": 0.23457232490181923}
|
| 176 |
+
{"time": "07:26:27", "event": "step", "step": 1600, "rows_seen": 51200, "lr": 1.4621466498270773e-05, "tok_per_s": 866.9900311792818, "loss": 0.5837186761200428, "entropy": 0.5831431988626719, "sharp": 0.208416230045259}
|
| 177 |
+
{"time": "07:27:50", "event": "val", "step": 1600, "rows_seen": 51200, "n": 500, "dropped_long": 0, "noul_ce": 0.5976080298423767, "noul_T": 0.8093874722626947, "noul_auc": 0.7340804374609712, "noul_acc": 0.712, "noul_f1": 0.7677419354838709, "noul_nll": 0.5953607472479343, "noul_ece": 0.08737396818399429, "noul_brier": 0.20371699830404988, "noul_sharp": 0.1539689004421234, "entropy": 0.6233376860618591}
|
| 178 |
+
{"time": "07:31:07", "event": "step", "step": 1610, "rows_seen": 51520, "lr": 1.4162319555398557e-05, "tok_per_s": 865.0027246097633, "loss": 0.5824661366641521, "entropy": 0.5616700779646635, "sharp": 0.2203763872385025}
|
| 179 |
+
{"time": "07:34:27", "event": "step", "step": 1620, "rows_seen": 51840, "lr": 1.3713226004782476e-05, "tok_per_s": 865.2113313324941, "loss": 0.5755713731050491, "entropy": 0.5716441906988621, "sharp": 0.2061484344303608}
|
| 180 |
+
{"time": "07:37:46", "event": "step", "step": 1630, "rows_seen": 52160, "lr": 1.3274303613422823e-05, "tok_per_s": 865.4375472902304, "loss": 0.4791920334100723, "entropy": 0.5218443982303143, "sharp": 0.2599113415926695}
|
| 181 |
+
{"time": "07:40:56", "event": "step", "step": 1640, "rows_seen": 52480, "lr": 1.284566748110963e-05, "tok_per_s": 865.6201273503268, "loss": 0.5344015574082732, "entropy": 0.5941757177934051, "sharp": 0.16892157308757305}
|
| 182 |
+
{"time": "07:44:14", "event": "step", "step": 1650, "rows_seen": 52800, "lr": 1.2427430010239793e-05, "tok_per_s": 865.8557646509853, "loss": 0.6305450797080994, "entropy": 0.5690765976905823, "sharp": 0.20026560313999653}
|
| 183 |
+
{"time": "07:47:27", "event": "step", "step": 1660, "rows_seen": 53120, "lr": 1.2019700876341466e-05, "tok_per_s": 866.0156838071066, "loss": 0.6006788238883018, "entropy": 0.6059622317552567, "sharp": 0.16662540659308434}
|
| 184 |
+
{"time": "07:50:45", "event": "step", "step": 1670, "rows_seen": 53440, "lr": 1.162258699931355e-05, "tok_per_s": 866.2215443997039, "loss": 0.5768307149410248, "entropy": 0.5523998439311981, "sharp": 0.2309844885021448}
|
| 185 |
+
{"time": "07:54:11", "event": "step", "step": 1680, "rows_seen": 53760, "lr": 1.1236192515387821e-05, "tok_per_s": 866.4895945164104, "loss": 0.5331007428467274, "entropy": 0.5487821996212006, "sharp": 0.2286573350429535}
|
| 186 |
+
{"time": "07:57:29", "event": "step", "step": 1690, "rows_seen": 54080, "lr": 1.0860618749820988e-05, "tok_per_s": 866.6409380504915, "loss": 0.6583492457866669, "entropy": 0.543205164372921, "sharp": 0.22740988060832024}
|
| 187 |
+
{"time": "08:00:51", "event": "step", "step": 1700, "rows_seen": 54400, "lr": 1.0495964190323887e-05, "tok_per_s": 866.8371741386654, "loss": 0.5540509819984436, "entropy": 0.5396690703928471, "sharp": 0.2391125001013279}
|
| 188 |
+
{"time": "08:02:14", "event": "val", "step": 1700, "rows_seen": 54400, "n": 500, "dropped_long": 0, "noul_ce": 0.6087808012962341, "noul_T": 0.8466512841060159, "noul_auc": 0.7369158326441748, "noul_acc": 0.688, "noul_f1": 0.7337883959044369, "noul_nll": 0.6042039534449577, "noul_ece": 0.10420507621765136, "noul_brier": 0.20765443068458195, "noul_sharp": 0.14992202818393707, "entropy": 0.627943754196167}
|
| 189 |
+
{"time": "08:05:27", "event": "step", "step": 1710, "rows_seen": 54720, "lr": 1.0142324461234822e-05, "tok_per_s": 864.9507188575059, "loss": 0.585735060274601, "entropy": 0.5785786509513855, "sharp": 0.198761984705925}
|
| 190 |
+
{"time": "08:08:35", "event": "step", "step": 1720, "rows_seen": 55040, "lr": 9.799792298443728e-06, "tok_per_s": 865.062459771275, "loss": 0.6163154430687428, "entropy": 0.6254133731126785, "sharp": 0.15129372477531433}
|
| 191 |
+
{"time": "08:11:56", "event": "step", "step": 1730, "rows_seen": 55360, "lr": 9.468457525073774e-06, "tok_per_s": 865.2612766702027, "loss": 0.5913186892867088, "entropy": 0.5937622636556625, "sharp": 0.17632815055549145}
|
| 192 |
+
{"time": "08:15:13", "event": "step", "step": 1740, "rows_seen": 55680, "lr": 9.148407027926849e-06, "tok_per_s": 865.474977554104, "loss": 0.5795738101005554, "entropy": 0.572651170194149, "sharp": 0.19926317408680916}
|
| 193 |
+
{"time": "08:18:30", "event": "step", "step": 1750, "rows_seen": 56000, "lr": 8.839724734698926e-06, "tok_per_s": 865.6713422897707, "loss": 0.6321916896849871, "entropy": 0.6034434624016285, "sharp": 0.17158264946192503}
|
| 194 |
+
{"time": "08:21:49", "event": "step", "step": 1760, "rows_seen": 56320, "lr": 8.542491591971554e-06, "tok_per_s": 865.870626560965, "loss": 0.5533383935689926, "entropy": 0.5550361387431622, "sharp": 0.22121554985642433}
|
| 195 |
+
{"time": "08:25:18", "event": "step", "step": 1770, "rows_seen": 56640, "lr": 8.256785543984993e-06, "tok_per_s": 866.0887779623931, "loss": 0.6392429247498512, "entropy": 0.5591999962925911, "sharp": 0.21259432472288609}
|
| 196 |
+
{"time": "08:28:40", "event": "step", "step": 1780, "rows_seen": 56960, "lr": 7.982681512198684e-06, "tok_per_s": 866.2762164464081, "loss": 0.5125716291368008, "entropy": 0.5612562373280525, "sharp": 0.21247396618127823}
|
| 197 |
+
{"time": "08:31:56", "event": "step", "step": 1790, "rows_seen": 57280, "lr": 7.720251375644373e-06, "tok_per_s": 866.452230026773, "loss": 0.6011955151334405, "entropy": 0.5853227786719799, "sharp": 0.2005887497216463}
|
| 198 |
+
{"time": "08:35:11", "event": "step", "step": 1800, "rows_seen": 57600, "lr": 7.469563952077155e-06, "tok_per_s": 866.5909812935364, "loss": 0.5528437308967113, "entropy": 0.5819540824741125, "sharp": 0.200270252302289}
|
| 199 |
+
{"time": "08:36:33", "event": "val", "step": 1800, "rows_seen": 57600, "n": 500, "dropped_long": 0, "noul_ce": 0.6052120923995972, "noul_T": 0.8145893504348751, "noul_auc": 0.7362744932574978, "noul_acc": 0.698, "noul_f1": 0.7487520798668885, "noul_nll": 0.6009124745875597, "noul_ece": 0.11032294648885728, "noul_brier": 0.20612982763447685, "noul_sharp": 0.14916689693927765, "entropy": 0.6288390159606934}
|
| 200 |
+
{"time": "08:39:55", "event": "step", "step": 1810, "rows_seen": 57920, "lr": 7.230684979929109e-06, "tok_per_s": 864.8446779772379, "loss": 0.6802630797028542, "entropy": 0.5762905273586512, "sharp": 0.19687986932694912}
|
| 201 |
+
{"time": "08:43:11", "event": "step", "step": 1820, "rows_seen": 58240, "lr": 7.003677101070616e-06, "tok_per_s": 865.0183260768623, "loss": 0.5486676767468452, "entropy": 0.5523110702633858, "sharp": 0.22993313148617744}
|
| 202 |
+
{"time": "08:46:33", "event": "step", "step": 1830, "rows_seen": 58560, "lr": 6.7885998443835666e-06, "tok_per_s": 865.2524033745451, "loss": 0.5589603260159492, "entropy": 0.571803405880928, "sharp": 0.20851874724030495}
|
| 203 |
+
{"time": "08:49:54", "event": "step", "step": 1840, "rows_seen": 58880, "lr": 6.585509610150972e-06, "tok_per_s": 865.425323878417, "loss": 0.6165741607546806, "entropy": 0.5315998122096062, "sharp": 0.26157208532094955}
|
| 204 |
+
{"time": "08:53:12", "event": "step", "step": 1850, "rows_seen": 59200, "lr": 6.394459655266939e-06, "tok_per_s": 865.6239979498104, "loss": 0.5750828757882118, "entropy": 0.5362860821187496, "sharp": 0.24811850674450397}
|
| 205 |
+
{"time": "08:56:29", "event": "step", "step": 1860, "rows_seen": 59520, "lr": 6.215500079271015e-06, "tok_per_s": 865.7975740674974, "loss": 0.6258493661880493, "entropy": 0.5759691651910543, "sharp": 0.20784255117177963}
|
| 206 |
+
{"time": "08:59:52", "event": "step", "step": 1870, "rows_seen": 59840, "lr": 6.048677811210399e-06, "tok_per_s": 866.0648945469927, "loss": 0.530236691236496, "entropy": 0.5640207231044769, "sharp": 0.21961429715156555}
|
| 207 |
+
{"time": "09:03:11", "event": "step", "step": 1880, "rows_seen": 60160, "lr": 5.894036597333646e-06, "tok_per_s": 866.2384742012358, "loss": 0.6218286715447903, "entropy": 0.5687127485871315, "sharp": 0.2157088192179799}
|
| 208 |
+
{"time": "09:06:30", "event": "step", "step": 1890, "rows_seen": 60480, "lr": 5.751616989618974e-06, "tok_per_s": 866.452715393574, "loss": 0.5751552283763885, "entropy": 0.5889202766120434, "sharp": 0.19346787314862013}
|
| 209 |
+
{"time": "09:09:47", "event": "step", "step": 1900, "rows_seen": 60800, "lr": 5.6214563351401916e-06, "tok_per_s": 866.6774859624592, "loss": 0.5640844162553549, "entropy": 0.5370340533554554, "sharp": 0.24669859185814857}
|
| 210 |
+
{"time": "09:11:10", "event": "val", "step": 1900, "rows_seen": 60800, "n": 500, "dropped_long": 0, "noul_ce": 0.6055179238319397, "noul_T": 0.8497129372853961, "noul_auc": 0.7349243050750198, "noul_acc": 0.706, "noul_f1": 0.7537688442211056, "noul_nll": 0.6011486613601447, "noul_ece": 0.10878034269809725, "noul_brier": 0.20633128804763895, "noul_sharp": 0.1527804285287857, "entropy": 0.6248840689659119}
|
| 211 |
+
{"time": "09:14:31", "event": "step", "step": 1910, "rows_seen": 61120, "lr": 5.5035887662730904e-06, "tok_per_s": 865.0618229322151, "loss": 0.5067151822149754, "entropy": 0.5295904539525509, "sharp": 0.24393040500581264}
|
| 212 |
+
{"time": "09:17:46", "event": "step", "step": 1920, "rows_seen": 61440, "lr": 5.398045191744809e-06, "tok_per_s": 865.2467262371333, "loss": 0.6202656850218773, "entropy": 0.5965881422162056, "sharp": 0.18369158543646336}
|
| 213 |
+
{"time": "09:21:05", "event": "step", "step": 1930, "rows_seen": 61760, "lr": 5.304853288528564e-06, "tok_per_s": 865.4552737597165, "loss": 0.5690196040086448, "entropy": 0.5801675412803888, "sharp": 0.18611296638846397}
|
| 214 |
+
{"time": "09:24:19", "event": "step", "step": 1940, "rows_seen": 62080, "lr": 5.224037494585838e-06, "tok_per_s": 865.5658108742517, "loss": 0.5749633349478245, "entropy": 0.5816404335200787, "sharp": 0.20370473712682724}
|
| 215 |
+
{"time": "09:27:37", "event": "step", "step": 1950, "rows_seen": 62400, "lr": 5.1556190024579555e-06, "tok_per_s": 865.7379075901347, "loss": 0.5325334705412388, "entropy": 0.5834958739578724, "sharp": 0.19024456664919853}
|
| 216 |
+
{"time": "09:30:51", "event": "step", "step": 1960, "rows_seen": 62720, "lr": 5.099615753708697e-06, "tok_per_s": 865.92474207002, "loss": 0.5755946114659309, "entropy": 0.5646393317729235, "sharp": 0.21146801952272654}
|
| 217 |
+
{"time": "09:34:06", "event": "step", "step": 1970, "rows_seen": 63040, "lr": 5.056042434219443e-06, "tok_per_s": 866.0472846601211, "loss": 0.5608649589121342, "entropy": 0.5673411414027214, "sharp": 0.1997239999473095}
|
| 218 |
+
{"time": "09:37:21", "event": "step", "step": 1980, "rows_seen": 63360, "lr": 5.0249104703380416e-06, "tok_per_s": 866.1877397765547, "loss": 0.6069850791245699, "entropy": 0.5908298445865512, "sharp": 0.19030447863042355}
|
| 219 |
+
{"time": "09:40:35", "event": "step", "step": 1990, "rows_seen": 63680, "lr": 5.006228025882471e-06, "tok_per_s": 866.3210320349252, "loss": 0.6116523742675781, "entropy": 0.5911118388175964, "sharp": 0.18360246531665325}
|
| 220 |
+
{"time": "09:43:57", "event": "step", "step": 2000, "rows_seen": 64000, "lr": 5e-06, "tok_per_s": 866.4750190096742, "loss": 0.5739720612764359, "entropy": 0.5838642567396164, "sharp": 0.19943976029753685}
|
| 221 |
+
{"time": "09:45:19", "event": "val", "step": 2000, "rows_seen": 64000, "n": 500, "dropped_long": 0, "noul_ce": 0.6036874651908875, "noul_T": 0.8261076472413659, "noul_auc": 0.734789286256772, "noul_acc": 0.704, "noul_f1": 0.7533333333333333, "noul_nll": 0.5996137060075998, "noul_ece": 0.09713243198394778, "noul_brier": 0.20562990025747735, "noul_sharp": 0.1509152203798294, "entropy": 0.6260700821876526}
|
| 222 |
+
{"time": "09:48:16", "event": "ckpt", "tag": "ckpt-64000", "rows_seen": 64000, "replay_seen": 3368, "step": 2000, "tokens": 35733211, "tok_per_s": 861.0871689455226, "rows_per_s": 1.623411912087117, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5864824652671814, "noul_T": 0.8028773328820736, "choice_ce": 0.6727158427238464, "choice_T": 0.9985050815668384, "score_ce": 1.0665595531463623, "score_T": 1.0349116888300849, "noul_auc": 0.7217706666666668, "noul_acc": 0.686, "noul_f1": 0.751188589540412, "noul_nll": 0.5977250373587012, "noul_ece": 0.08486032821238039, "noul_brier": 0.2051691787868441, "noul_sharp": 0.1511927843093872, "entropy": 0.6252002716064453}, "dropped_long": 0, "peak_gib": 18.235705852508545}
|
meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"v_proj",
|
| 39 |
+
"out_proj",
|
| 40 |
+
"down_proj",
|
| 41 |
+
"up_proj",
|
| 42 |
+
"in_proj_z",
|
| 43 |
+
"gate_proj",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"q_proj",
|
| 46 |
+
"k_proj",
|
| 47 |
+
"in_proj_qkv"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0002604858891572803,
|
| 4 |
+
-0.00026073516346514225,
|
| 5 |
+
-0.0031090632546693087,
|
| 6 |
+
-0.0014582076109945774,
|
| 7 |
+
-0.0001874673762358725,
|
| 8 |
+
0.0020599483978003263,
|
| 9 |
+
0.0011498567182570696,
|
| 10 |
+
-0.0025537230540066957,
|
| 11 |
+
-0.0009152532438747585,
|
| 12 |
+
-0.000312424439471215,
|
| 13 |
+
-0.0007771191885694861,
|
| 14 |
+
-0.0001489325222792104,
|
| 15 |
+
-0.0006520861061289907,
|
| 16 |
+
-0.0018110942328348756,
|
| 17 |
+
-0.0030925078317523003,
|
| 18 |
+
-0.005080677568912506,
|
| 19 |
+
-0.005937784910202026,
|
| 20 |
+
-0.0043688504956662655,
|
| 21 |
+
-0.004944807384163141,
|
| 22 |
+
-0.004981663543730974,
|
| 23 |
+
-0.0040712254121899605,
|
| 24 |
+
-0.0034084897488355637,
|
| 25 |
+
-0.005614522844552994,
|
| 26 |
+
-0.0034949742257595062
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 1.0112745115984856,
|
| 5 |
+
"choice": 0.9566453363997803,
|
| 6 |
+
"score": 0.9510701603936809
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 2087,
|
| 11 |
+
"T": 1.0112745115984856,
|
| 12 |
+
"ce_before": 0.582038938999176,
|
| 13 |
+
"ce_after": 0.5820286870002747
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 0.9566453363997803,
|
| 18 |
+
"ce_before": 0.6593058705329895,
|
| 19 |
+
"ce_after": 0.6588473916053772
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 0.9510701603936809,
|
| 24 |
+
"ce_before": 1.0452001094818115,
|
| 25 |
+
"ce_after": 1.0444562435150146
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-F2syn4-rp/ckpt-4000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-F2syn4-rp/ckpt-4000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-F2syn4-rp/ckpt-4000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e15/F2-syn-f2_wide_s0",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "F2-syn",
|
| 108 |
+
"name": "F2-syn-f2_wide_s0",
|
| 109 |
+
"train_sha256": "bacf92d9a3b99b17963cf060a6c6696da5a891b0e2097c0c13e4ef231cad8bfd",
|
| 110 |
+
"n_train": 64000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 125 steps",
|
| 128 |
+
"total_rows": 4000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-4000",
|
| 137 |
+
"rows_seen": 4000,
|
| 138 |
+
"replay_seen": 210,
|
| 139 |
+
"step": 125,
|
| 140 |
+
"tokens": 2304188,
|
| 141 |
+
"tok_per_s": 755.6208975504845,
|
| 142 |
+
"rows_per_s": 1.3806008782482386,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 2200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.582038938999176,
|
| 147 |
+
"noul_T": 1.0112745115984856,
|
| 148 |
+
"choice_ce": 0.6593058705329895,
|
| 149 |
+
"choice_T": 0.9566453363997803,
|
| 150 |
+
"score_ce": 1.0452001094818115,
|
| 151 |
+
"score_T": 0.9510701603936809,
|
| 152 |
+
"noul_auc": 0.7029088000000001,
|
| 153 |
+
"noul_acc": 0.6685,
|
| 154 |
+
"noul_f1": 0.7539888682745826,
|
| 155 |
+
"noul_nll": 0.6000991474576295,
|
| 156 |
+
"noul_ece": 0.028662465870380415,
|
| 157 |
+
"noul_brier": 0.2069418974020736,
|
| 158 |
+
"noul_sharp": 0.1901477426290512,
|
| 159 |
+
"entropy": 0.5843608379364014
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-F2syn4-rp/train_log.jsonl
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "03:40:57", "event": "start", "data": "data/e15/F2-syn-f2_wide_s0", "data_manifest": {"arm": "F2-syn", "name": "F2-syn-f2_wide_s0", "train_sha256": "bacf92d9a3b99b17963cf060a6c6696da5a891b0e2097c0c13e4ef231cad8bfd", "n_train": 64000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 125 steps", "total_rows": 4000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 278.40041422843933}
|
| 2 |
+
{"time": "03:44:34", "event": "step", "step": 10, "rows_seen": 320, "lr": 9.923040269974204e-05, "tok_per_s": 853.5862749683131, "loss": 0.5829928889870644, "entropy": 0.6182952374219894, "sharp": 0.1685207523405552}
|
| 3 |
+
{"time": "03:48:13", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.55208574963224e-05, "tok_per_s": 869.5953326423811, "loss": 0.5975874159485102, "entropy": 0.6500829588621855, "sharp": 0.10333448555320501}
|
| 4 |
+
{"time": "03:51:39", "event": "step", "step": 30, "rows_seen": 960, "lr": 8.897432118673851e-05, "tok_per_s": 874.9070767988629, "loss": 0.6620418652892113, "entropy": 0.6581599563360214, "sharp": 0.10284903552383184}
|
| 5 |
+
{"time": "03:55:09", "event": "step", "step": 40, "rows_seen": 1280, "lr": 8.002250216471678e-05, "tok_per_s": 877.7364415378707, "loss": 0.6499629803001881, "entropy": 0.5313977859914303, "sharp": 0.24373930133879185}
|
| 6 |
+
{"time": "03:58:41", "event": "step", "step": 50, "rows_seen": 1600, "lr": 6.925572411108201e-05, "tok_per_s": 878.7957394375853, "loss": 0.6206380501389503, "entropy": 0.6030138731002808, "sharp": 0.1754997679963708}
|
| 7 |
+
{"time": "04:01:58", "event": "step", "step": 60, "rows_seen": 1920, "lr": 5.738399736402462e-05, "tok_per_s": 878.7219302956797, "loss": 0.5693474039435387, "entropy": 0.5707865208387375, "sharp": 0.21803556755185127}
|
| 8 |
+
{"time": "04:05:24", "event": "step", "step": 70, "rows_seen": 2240, "lr": 4.519019760655705e-05, "tok_per_s": 881.1888057171714, "loss": 0.5648954659700394, "entropy": 0.6217266377061605, "sharp": 0.1480824314057827}
|
| 9 |
+
{"time": "04:08:59", "event": "step", "step": 80, "rows_seen": 2560, "lr": 3.3478439481564835e-05, "tok_per_s": 881.0109491008251, "loss": 0.5980550739914179, "entropy": 0.6094800420105457, "sharp": 0.15593381505459547}
|
| 10 |
+
{"time": "04:12:31", "event": "step", "step": 90, "rows_seen": 2880, "lr": 2.3021049613197416e-05, "tok_per_s": 882.1623330147426, "loss": 0.5693720008712262, "entropy": 0.5007136785425246, "sharp": 0.2730809859931469}
|
| 11 |
+
{"time": "04:16:02", "event": "step", "step": 100, "rows_seen": 3200, "lr": 1.4507635874863332e-05, "tok_per_s": 882.4924567742866, "loss": 0.5537372324615717, "entropy": 0.5923325587064028, "sharp": 0.18620688188821077}
|
| 12 |
+
{"time": "04:17:27", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.6002792716026306, "noul_T": 1.140800174683004, "noul_auc": 0.7158242586124237, "noul_acc": 0.68, "noul_f1": 0.7523219814241486, "noul_nll": 0.5969109509661794, "noul_ece": 0.06848560547828675, "noul_brier": 0.20548950074612618, "noul_sharp": 0.20036578178405762, "entropy": 0.5774202346801758}
|
| 13 |
+
{"time": "04:20:51", "event": "step", "step": 110, "rows_seen": 3520, "lr": 8.49961150805399e-06, "tok_per_s": 851.1136948342781, "loss": 0.6563278790563345, "entropy": 0.5875902026891708, "sharp": 0.17992982640862465}
|
| 14 |
+
{"time": "04:24:18", "event": "step", "step": 120, "rows_seen": 3840, "lr": 5.3931729785351795e-06, "tok_per_s": 854.0715355621543, "loss": 0.5456394553184509, "entropy": 0.5497702546417713, "sharp": 0.2220058124512434}
|
| 15 |
+
{"time": "04:31:48", "event": "ckpt", "tag": "ckpt-4000", "rows_seen": 4000, "replay_seen": 210, "step": 125, "tokens": 2304188, "tok_per_s": 755.6208975504845, "rows_per_s": 1.3806008782482386, "val": {"n": 2200, "dropped_long": 0, "noul_ce": 0.582038938999176, "noul_T": 1.0112745115984856, "choice_ce": 0.6593058705329895, "choice_T": 0.9566453363997803, "score_ce": 1.0452001094818115, "score_T": 0.9510701603936809, "noul_auc": 0.7029088000000001, "noul_acc": 0.6685, "noul_f1": 0.7539888682745826, "noul_nll": 0.6000991474576295, "noul_ece": 0.028662465870380415, "noul_brier": 0.2069418974020736, "noul_sharp": 0.1901477426290512, "entropy": 0.5843608379364014}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"out_proj",
|
| 39 |
+
"o_proj",
|
| 40 |
+
"gate_proj",
|
| 41 |
+
"q_proj",
|
| 42 |
+
"in_proj_z",
|
| 43 |
+
"in_proj_qkv",
|
| 44 |
+
"down_proj",
|
| 45 |
+
"v_proj",
|
| 46 |
+
"up_proj",
|
| 47 |
+
"k_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.00043673961772583425,
|
| 4 |
+
-0.000437002134276554,
|
| 5 |
+
-0.0024751147720962763,
|
| 6 |
+
-0.0020742828492075205,
|
| 7 |
+
-0.0005924999713897705,
|
| 8 |
+
0.0018874861998483539,
|
| 9 |
+
0.0016518369084224105,
|
| 10 |
+
-0.00206023920327425,
|
| 11 |
+
-7.941335934447125e-05,
|
| 12 |
+
-0.00025902234483510256,
|
| 13 |
+
0.00028893505805172026,
|
| 14 |
+
0.0005586668266914785,
|
| 15 |
+
1.6555903130210936e-05,
|
| 16 |
+
-0.0037094794679433107,
|
| 17 |
+
-0.004494857974350452,
|
| 18 |
+
-0.005148923024535179,
|
| 19 |
+
-0.0059170350432395935,
|
| 20 |
+
-0.004471797030419111,
|
| 21 |
+
-0.00495561771094799,
|
| 22 |
+
-0.0050347549840807915,
|
| 23 |
+
-0.0038398555479943752,
|
| 24 |
+
-0.0034546656534075737,
|
| 25 |
+
-0.005773663055151701,
|
| 26 |
+
-0.0035197518300265074
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8668834258839889,
|
| 5 |
+
"choice": 1.0474929177562735,
|
| 6 |
+
"score": 0.9844687264725452
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8668834258839889,
|
| 12 |
+
"ce_before": 0.5594630241394043,
|
| 13 |
+
"ce_after": 0.5578091144561768
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 1.0474929177562735,
|
| 18 |
+
"ce_before": 0.6948021054267883,
|
| 19 |
+
"ce_after": 0.6943069100379944
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 0.9844687264725452,
|
| 24 |
+
"ce_before": 1.0638198852539062,
|
| 25 |
+
"ce_after": 1.0637402534484863
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-G-d4096-n16k-syn/ckpt-16000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/ckpt-16000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-G-d4096-n16k-syn/ckpt-16000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e18/G-d4096-n16k",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "G-d4096-n16k",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "4b56315ac44284be6d58d7deac9c2c7ba1fc6dfa0443f1ae5d267fad7757371c",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
|
| 128 |
+
"total_rows": 16000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-16000",
|
| 137 |
+
"rows_seen": 16000,
|
| 138 |
+
"replay_seen": 842,
|
| 139 |
+
"step": 500,
|
| 140 |
+
"tokens": 9137216,
|
| 141 |
+
"tok_per_s": 835.462918625799,
|
| 142 |
+
"rows_per_s": 1.539951170025349,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5594630241394043,
|
| 147 |
+
"noul_T": 0.8668834258839889,
|
| 148 |
+
"choice_ce": 0.6948021054267883,
|
| 149 |
+
"choice_T": 1.0474929177562735,
|
| 150 |
+
"score_ce": 1.0638198852539062,
|
| 151 |
+
"score_T": 0.9844687264725452,
|
| 152 |
+
"noul_auc": 0.7413290666666665,
|
| 153 |
+
"noul_acc": 0.695,
|
| 154 |
+
"noul_f1": 0.7677075399847677,
|
| 155 |
+
"noul_nll": 0.5746213684976101,
|
| 156 |
+
"noul_ece": 0.048617555305361757,
|
| 157 |
+
"noul_brier": 0.19619758724711828,
|
| 158 |
+
"noul_sharp": 0.18243390321731567,
|
| 159 |
+
"entropy": 0.5923757553100586
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-G-d4096-n16k-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "06:36:01", "event": "start", "data": "data/e18/G-d4096-n16k", "data_manifest": {"arm": "G-d4096-n16k", "name": null, "train_sha256": "4b56315ac44284be6d58d7deac9c2c7ba1fc6dfa0443f1ae5d267fad7757371c", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 80.52906060218811}
|
| 2 |
+
{"time": "06:39:26", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 860.4663116721329, "loss": 0.6019687056541443, "entropy": 0.6530313715338707, "sharp": 0.1172491256147623}
|
| 3 |
+
{"time": "06:42:53", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 869.4883319675097, "loss": 0.6082677915692329, "entropy": 0.6271356660872698, "sharp": 0.15409039799124002}
|
| 4 |
+
{"time": "06:46:22", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 877.9122720016684, "loss": 0.6429466586560011, "entropy": 0.6207066047936678, "sharp": 0.16099937725812197}
|
| 5 |
+
{"time": "06:49:52", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 875.6979383788292, "loss": 0.60123236104846, "entropy": 0.5957193709909916, "sharp": 0.19625727739185095}
|
| 6 |
+
{"time": "06:53:22", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 877.9517214882385, "loss": 0.6294720396399498, "entropy": 0.6109470501542091, "sharp": 0.1733695399016142}
|
| 7 |
+
{"time": "06:56:43", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 879.6811055552294, "loss": 0.6080367900431156, "entropy": 0.616622868925333, "sharp": 0.15778671763837337}
|
| 8 |
+
{"time": "07:00:03", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 881.2637199827609, "loss": 0.6042112521827221, "entropy": 0.616423737257719, "sharp": 0.16625128872692585}
|
| 9 |
+
{"time": "07:03:27", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 880.7753024816587, "loss": 0.5941497050225735, "entropy": 0.6047854125499725, "sharp": 0.17890577390789986}
|
| 10 |
+
{"time": "07:06:47", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 880.6433115212006, "loss": 0.6117764860391617, "entropy": 0.6087187211960554, "sharp": 0.18156814947724342}
|
| 11 |
+
{"time": "07:10:06", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 880.1072979384887, "loss": 0.582784254103899, "entropy": 0.6220661997795105, "sharp": 0.16008947975933552}
|
| 12 |
+
{"time": "07:11:29", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5812910795211792, "noul_T": 0.7417187366100655, "noul_auc": 0.74324483974954, "noul_acc": 0.704, "noul_f1": 0.7708978328173375, "noul_nll": 0.5833777135312557, "noul_ece": 0.07091195756196976, "noul_brier": 0.199204913071057, "noul_sharp": 0.15617571771144867, "entropy": 0.6190448999404907}
|
| 13 |
+
{"time": "07:15:02", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 850.1706548554492, "loss": 0.5881851203739643, "entropy": 0.5967134274542332, "sharp": 0.17742653656750917}
|
| 14 |
+
{"time": "07:18:38", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 853.1883262416984, "loss": 0.591743653640151, "entropy": 0.6224749013781548, "sharp": 0.1569033907726407}
|
| 15 |
+
{"time": "07:22:11", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 855.3329218283026, "loss": 0.6061377786099911, "entropy": 0.6265431679785252, "sharp": 0.14779445063322783}
|
| 16 |
+
{"time": "07:25:43", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 857.5362467564697, "loss": 0.5914337541908026, "entropy": 0.5883870637044311, "sharp": 0.18891076184809208}
|
| 17 |
+
{"time": "07:29:01", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 858.4715901132216, "loss": 0.6465858146548271, "entropy": 0.6004343405365944, "sharp": 0.18271316587924957}
|
| 18 |
+
{"time": "07:32:29", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 860.1715622446063, "loss": 0.6757050529122353, "entropy": 0.6000629663467407, "sharp": 0.1881898231804371}
|
| 19 |
+
{"time": "07:35:54", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 861.3615510754703, "loss": 0.6518629677593708, "entropy": 0.5970906540751457, "sharp": 0.18546851351857185}
|
| 20 |
+
{"time": "07:39:19", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 862.1109374350823, "loss": 0.5738454759120941, "entropy": 0.583740372210741, "sharp": 0.18634894769638777}
|
| 21 |
+
{"time": "07:42:54", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 862.7558560145757, "loss": 0.5746322870254517, "entropy": 0.6003875881433487, "sharp": 0.17547284066677094}
|
| 22 |
+
{"time": "07:46:17", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 863.2600051642689, "loss": 0.58967300131917, "entropy": 0.6155922915786505, "sharp": 0.1640684688463807}
|
| 23 |
+
{"time": "07:47:40", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5860417485237122, "noul_T": 0.7424384045195787, "noul_auc": 0.745962093466777, "noul_acc": 0.698, "noul_f1": 0.7544715447154472, "noul_nll": 0.5863566996008158, "noul_ece": 0.08091064447164535, "noul_brier": 0.2003083443752785, "noul_sharp": 0.15349411964416504, "entropy": 0.6227638721466064}
|
| 24 |
+
{"time": "07:51:11", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 848.6767212722873, "loss": 0.6390631757676601, "entropy": 0.5979266427457333, "sharp": 0.19080301746726036}
|
| 25 |
+
{"time": "07:54:44", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 850.0795628756969, "loss": 0.6112649329006672, "entropy": 0.613115580752492, "sharp": 0.16590545419603586}
|
| 26 |
+
{"time": "07:58:05", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 851.0855351817594, "loss": 0.6322213131934404, "entropy": 0.5826841779053211, "sharp": 0.20607813633978367}
|
| 27 |
+
{"time": "08:01:34", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 852.3971608079495, "loss": 0.605877798050642, "entropy": 0.6114602312445641, "sharp": 0.16823562514036894}
|
| 28 |
+
{"time": "08:05:03", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 853.5091906846634, "loss": 0.5960135720670223, "entropy": 0.5718173310160637, "sharp": 0.19682771619409323}
|
| 29 |
+
{"time": "08:08:20", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 854.0504285088928, "loss": 0.579134676605463, "entropy": 0.5985534302890301, "sharp": 0.17056022677570581}
|
| 30 |
+
{"time": "08:11:41", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 855.0010287191643, "loss": 0.5732387229800224, "entropy": 0.6161045879125595, "sharp": 0.16193007118999958}
|
| 31 |
+
{"time": "08:15:14", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 856.3287584232353, "loss": 0.639550156891346, "entropy": 0.6081734672188759, "sharp": 0.17208983469754457}
|
| 32 |
+
{"time": "08:18:48", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 857.2103251456322, "loss": 0.5954747684299946, "entropy": 0.5814548246562481, "sharp": 0.18849235959351063}
|
| 33 |
+
{"time": "08:22:23", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 858.3912481397809, "loss": 0.6369517706334591, "entropy": 0.5966799110174179, "sharp": 0.17582474276423454}
|
| 34 |
+
{"time": "08:23:45", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5713485479354858, "noul_T": 0.9142534388712987, "noul_auc": 0.7472785269446929, "noul_acc": 0.7, "noul_f1": 0.7713414634146342, "noul_nll": 0.575156541571021, "noul_ece": 0.07330047130584719, "noul_brier": 0.19594645452938084, "noul_sharp": 0.18920540809631348, "entropy": 0.5860469937324524}
|
| 35 |
+
{"time": "08:27:04", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 848.7079639131221, "loss": 0.5857135877013206, "entropy": 0.6141913942992687, "sharp": 0.15854477602988482}
|
| 36 |
+
{"time": "08:30:20", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 849.7106958674167, "loss": 0.6283030621707439, "entropy": 0.6059871464967728, "sharp": 0.16288513410836458}
|
| 37 |
+
{"time": "08:33:52", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 851.007084122665, "loss": 0.593388931825757, "entropy": 0.5887055024504662, "sharp": 0.18707896396517754}
|
| 38 |
+
{"time": "08:37:17", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 851.845477250502, "loss": 0.6230612844228745, "entropy": 0.592638436704874, "sharp": 0.17973813507705927}
|
| 39 |
+
{"time": "08:40:47", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 852.8969524018004, "loss": 0.580170426517725, "entropy": 0.5790281444787979, "sharp": 0.18796169757843018}
|
| 40 |
+
{"time": "08:44:06", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 853.4430106493527, "loss": 0.6324121579527855, "entropy": 0.5997336357831955, "sharp": 0.17613977938890457}
|
| 41 |
+
{"time": "08:47:26", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 854.0998189076893, "loss": 0.6068278234452009, "entropy": 0.5965613275766373, "sharp": 0.18619749415665865}
|
| 42 |
+
{"time": "08:50:45", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 854.8714660136978, "loss": 0.5972454128786922, "entropy": 0.5813715159893036, "sharp": 0.2005388792604208}
|
| 43 |
+
{"time": "08:54:12", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 855.7684070387056, "loss": 0.5637623192742467, "entropy": 0.5549548231065273, "sharp": 0.22489557601511478}
|
| 44 |
+
{"time": "08:57:44", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 856.4165320809252, "loss": 0.5802994929254055, "entropy": 0.5960481911897659, "sharp": 0.18519839458167553}
|
| 45 |
+
{"time": "08:59:07", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5756452083587646, "noul_T": 0.8978204390003309, "noul_auc": 0.7494050733320956, "noul_acc": 0.704, "noul_f1": 0.765079365079365, "noul_nll": 0.5774106721431017, "noul_ece": 0.05126348590850831, "noul_brier": 0.19678551376903097, "noul_sharp": 0.18214792013168335, "entropy": 0.5924673676490784}
|
| 46 |
+
{"time": "09:02:46", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 848.950374642579, "loss": 0.6325492691248655, "entropy": 0.6134136840701103, "sharp": 0.15889494959264994}
|
| 47 |
+
{"time": "09:06:18", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 849.9530896313898, "loss": 0.6160378735512495, "entropy": 0.6029532179236412, "sharp": 0.18491806276142597}
|
| 48 |
+
{"time": "09:09:55", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 851.0257814634969, "loss": 0.6465812809765339, "entropy": 0.6200941726565361, "sharp": 0.15572707168757915}
|
| 49 |
+
{"time": "09:13:25", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 851.7160850671631, "loss": 0.5824000854045153, "entropy": 0.5652601569890976, "sharp": 0.21601914800703526}
|
| 50 |
+
{"time": "09:16:51", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 852.6133150797151, "loss": 0.5791643224656582, "entropy": 0.5866317301988602, "sharp": 0.19250667095184326}
|
| 51 |
+
{"time": "09:20:23", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 853.454423899262, "loss": 0.5699873678386211, "entropy": 0.5872764643281698, "sharp": 0.1856958419084549}
|
| 52 |
+
{"time": "09:23:44", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 853.7884600877803, "loss": 0.6410699393600225, "entropy": 0.5948164910078049, "sharp": 0.18491088319569826}
|
| 53 |
+
{"time": "09:27:12", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 854.4142465987834, "loss": 0.6067685959860682, "entropy": 0.5829725377261639, "sharp": 0.19912430550903082}
|
| 54 |
+
{"time": "09:30:40", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 855.206303160224, "loss": 0.6164331883192062, "entropy": 0.6271679997444153, "sharp": 0.1476202653720975}
|
| 55 |
+
{"time": "09:33:59", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 855.6558373507921, "loss": 0.5958688221871853, "entropy": 0.6007783375680447, "sharp": 0.1774207465350628}
|
| 56 |
+
{"time": "09:35:22", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.573348879814148, "noul_T": 0.8806576974759365, "noul_auc": 0.7488987527636666, "noul_acc": 0.69, "noul_f1": 0.7527910685805422, "noul_nll": 0.5765451846420765, "noul_ece": 0.06672371339797975, "noul_brier": 0.1966955598773074, "noul_sharp": 0.18101058900356293, "entropy": 0.592964768409729}
|
| 57 |
+
{"time": "09:39:24", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9137216, "tok_per_s": 835.462918625799, "rows_per_s": 1.539951170025349, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5594630241394043, "noul_T": 0.8668834258839889, "choice_ce": 0.6948021054267883, "choice_T": 1.0474929177562735, "score_ce": 1.0638198852539062, "score_T": 0.9844687264725452, "noul_auc": 0.7413290666666665, "noul_acc": 0.695, "noul_f1": 0.7677075399847677, "noul_nll": 0.5746213684976101, "noul_ece": 0.048617555305361757, "noul_brier": 0.19619758724711828, "noul_sharp": 0.18243390321731567, "entropy": 0.5923757553100586}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"k_proj",
|
| 39 |
+
"down_proj",
|
| 40 |
+
"gate_proj",
|
| 41 |
+
"in_proj_qkv",
|
| 42 |
+
"q_proj",
|
| 43 |
+
"in_proj_z",
|
| 44 |
+
"up_proj",
|
| 45 |
+
"v_proj",
|
| 46 |
+
"o_proj",
|
| 47 |
+
"out_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0003904224431607872,
|
| 4 |
+
-0.0003906911297235638,
|
| 5 |
+
-0.0027016778476536274,
|
| 6 |
+
-0.001719787484034896,
|
| 7 |
+
-0.0004692776419688016,
|
| 8 |
+
0.002059800550341606,
|
| 9 |
+
0.0012747086584568024,
|
| 10 |
+
-0.0028597551863640547,
|
| 11 |
+
-0.000726852857042104,
|
| 12 |
+
-0.0005554182571358979,
|
| 13 |
+
-0.0002128184278262779,
|
| 14 |
+
9.807036985876039e-05,
|
| 15 |
+
-0.0002962547878269106,
|
| 16 |
+
-0.004006666596978903,
|
| 17 |
+
-0.004581575747579336,
|
| 18 |
+
-0.005236349999904633,
|
| 19 |
+
-0.006067482754588127,
|
| 20 |
+
-0.00441952608525753,
|
| 21 |
+
-0.004967718385159969,
|
| 22 |
+
-0.00510936276987195,
|
| 23 |
+
-0.004137770738452673,
|
| 24 |
+
-0.003533437615260482,
|
| 25 |
+
-0.005713726859539747,
|
| 26 |
+
-0.003560874378308654
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8614883118654615,
|
| 5 |
+
"choice": 1.0608593759639455,
|
| 6 |
+
"score": 1.0216659864400268
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8614883118654615,
|
| 12 |
+
"ce_before": 0.5651941895484924,
|
| 13 |
+
"ce_after": 0.5634161233901978
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 1.0608593759639455,
|
| 18 |
+
"ce_before": 0.6922976970672607,
|
| 19 |
+
"ce_after": 0.6915320754051208
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 1.0216659864400268,
|
| 24 |
+
"ce_before": 1.0891292095184326,
|
| 25 |
+
"ce_after": 1.088997483253479
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-G-d512-n16k-a025-syn/ckpt-16000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/ckpt-16000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-G-d512-n16k-a025-syn/ckpt-16000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e18/G-d512-n16k-a025",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "G-d512-n16k-a025",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "dfc33198cc659ed918234b1458803d6b16d0a6a7bf048b62788018b1e7afda22",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
|
| 128 |
+
"total_rows": 16000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-16000",
|
| 137 |
+
"rows_seen": 16000,
|
| 138 |
+
"replay_seen": 842,
|
| 139 |
+
"step": 500,
|
| 140 |
+
"tokens": 9113372,
|
| 141 |
+
"tok_per_s": 836.8005348893143,
|
| 142 |
+
"rows_per_s": 1.5464522468215414,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5651941895484924,
|
| 147 |
+
"noul_T": 0.8614883118654615,
|
| 148 |
+
"choice_ce": 0.6922976970672607,
|
| 149 |
+
"choice_T": 1.0608593759639455,
|
| 150 |
+
"score_ce": 1.0891292095184326,
|
| 151 |
+
"score_T": 1.0216659864400268,
|
| 152 |
+
"noul_auc": 0.7393109333333333,
|
| 153 |
+
"noul_acc": 0.695,
|
| 154 |
+
"noul_f1": 0.7648419429452583,
|
| 155 |
+
"noul_nll": 0.5775874657034874,
|
| 156 |
+
"noul_ece": 0.05573579064011575,
|
| 157 |
+
"noul_brier": 0.19712157424601043,
|
| 158 |
+
"noul_sharp": 0.17942030727863312,
|
| 159 |
+
"entropy": 0.5986356139183044
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a025-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "11:51:26", "event": "start", "data": "data/e18/G-d512-n16k-a025", "data_manifest": {"arm": "G-d512-n16k-a025", "name": null, "train_sha256": "dfc33198cc659ed918234b1458803d6b16d0a6a7bf048b62788018b1e7afda22", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 102.44699573516846}
|
| 2 |
+
{"time": "11:54:55", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 864.7916055502311, "loss": 0.6066994518041611, "entropy": 0.6412589959800243, "sharp": 0.13278362713754177}
|
| 3 |
+
{"time": "11:58:22", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 869.9417254957506, "loss": 0.6137999035418034, "entropy": 0.6072024777531624, "sharp": 0.18162340950220823}
|
| 4 |
+
{"time": "12:01:52", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 872.9336615920942, "loss": 0.608553022146225, "entropy": 0.5837096571922302, "sharp": 0.19932523928582668}
|
| 5 |
+
{"time": "12:05:15", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 874.8300840785947, "loss": 0.5956386290490627, "entropy": 0.5935255698859692, "sharp": 0.1859582206234336}
|
| 6 |
+
{"time": "12:08:45", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 877.1826768232614, "loss": 0.5792789123952389, "entropy": 0.5912551339715719, "sharp": 0.19122331030666828}
|
| 7 |
+
{"time": "12:12:07", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 879.1711437651193, "loss": 0.591133963316679, "entropy": 0.6127467676997185, "sharp": 0.17640301305800676}
|
| 8 |
+
{"time": "12:15:36", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 879.4161911098678, "loss": 0.5903284549713135, "entropy": 0.6126270703971386, "sharp": 0.17047567758709192}
|
| 9 |
+
{"time": "12:19:01", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 879.7871098181879, "loss": 0.6167854014784098, "entropy": 0.6174231674522161, "sharp": 0.16632107831537724}
|
| 10 |
+
{"time": "12:22:32", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 881.2826350863772, "loss": 0.6389051303267479, "entropy": 0.625249695032835, "sharp": 0.1464562676846981}
|
| 11 |
+
{"time": "12:26:00", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 881.3385289942395, "loss": 0.5815365370362997, "entropy": 0.6120374090969563, "sharp": 0.16943395882844925}
|
| 12 |
+
{"time": "12:27:23", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5814564228057861, "noul_T": 0.8672998241595238, "noul_auc": 0.7443924997046464, "noul_acc": 0.69, "noul_f1": 0.7527910685805422, "noul_nll": 0.5825258432626724, "noul_ece": 0.05298670160770416, "noul_brier": 0.19882718893589732, "noul_sharp": 0.1751517355442047, "entropy": 0.6024430394172668}
|
| 13 |
+
{"time": "12:30:46", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 851.3404078122533, "loss": 0.6237065345048904, "entropy": 0.6054281741380692, "sharp": 0.16873749345541}
|
| 14 |
+
{"time": "12:34:22", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 854.7885817354835, "loss": 0.5789002180099487, "entropy": 0.6168119329959154, "sharp": 0.1597161963582039}
|
| 15 |
+
{"time": "12:37:58", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 857.8644964989522, "loss": 0.573807206004858, "entropy": 0.5982401445508003, "sharp": 0.18262134678661823}
|
| 16 |
+
{"time": "12:41:25", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 859.3548097428502, "loss": 0.6224136594682932, "entropy": 0.6134730502963066, "sharp": 0.16601629368960857}
|
| 17 |
+
{"time": "12:44:56", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 860.7887382893355, "loss": 0.6377713419497013, "entropy": 0.6314211450517178, "sharp": 0.1560398880392313}
|
| 18 |
+
{"time": "12:48:18", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 862.1350967429241, "loss": 0.632047925144434, "entropy": 0.5857958234846592, "sharp": 0.1888828594237566}
|
| 19 |
+
{"time": "12:51:48", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 863.4718705163386, "loss": 0.6244191564619541, "entropy": 0.6305734477937222, "sharp": 0.14076693169772625}
|
| 20 |
+
{"time": "12:55:14", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 864.7708170283213, "loss": 0.5865814350545406, "entropy": 0.5915018059313297, "sharp": 0.1897669155150652}
|
| 21 |
+
{"time": "12:58:41", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 865.5004189427955, "loss": 0.6329965218901634, "entropy": 0.6098176762461662, "sharp": 0.17066353932023048}
|
| 22 |
+
{"time": "13:02:08", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 866.4678965835312, "loss": 0.6042630039155483, "entropy": 0.6163564473390579, "sharp": 0.1707962043583393}
|
| 23 |
+
{"time": "13:03:31", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.582727313041687, "noul_T": 0.7897727166313554, "noul_auc": 0.7470422440127593, "noul_acc": 0.704, "noul_f1": 0.765079365079365, "noul_nll": 0.5816352546066046, "noul_ece": 0.057403193235397346, "noul_brier": 0.19800809253603252, "noul_sharp": 0.16509106755256653, "entropy": 0.6147062182426453}
|
| 24 |
+
{"time": "13:07:02", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 851.6907324857633, "loss": 0.6076771765947342, "entropy": 0.574773795902729, "sharp": 0.20614924654364586}
|
| 25 |
+
{"time": "13:10:25", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 852.9055286844496, "loss": 0.6054600290954113, "entropy": 0.5724468417465687, "sharp": 0.20186365209519863}
|
| 26 |
+
{"time": "13:13:45", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 854.1207840903256, "loss": 0.6015157401561737, "entropy": 0.5948246717453003, "sharp": 0.19528712332248688}
|
| 27 |
+
{"time": "13:17:10", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 855.2437696677423, "loss": 0.6041692979633808, "entropy": 0.6012048721313477, "sharp": 0.1831300426274538}
|
| 28 |
+
{"time": "13:20:49", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 856.2441869795409, "loss": 0.6232908889651299, "entropy": 0.6316490881145, "sharp": 0.14192802272737026}
|
| 29 |
+
{"time": "13:24:02", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 857.2126060603438, "loss": 0.5933821126818657, "entropy": 0.5947095453739166, "sharp": 0.17658598255366087}
|
| 30 |
+
{"time": "13:27:28", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 858.1744204107997, "loss": 0.6336482064798474, "entropy": 0.6361192837357521, "sharp": 0.13717816025018692}
|
| 31 |
+
{"time": "13:30:51", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 859.2125034619551, "loss": 0.6450212989002466, "entropy": 0.6063980869948864, "sharp": 0.17501452937722206}
|
| 32 |
+
{"time": "13:34:21", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 860.2619964805019, "loss": 0.6268995553255081, "entropy": 0.599021028727293, "sharp": 0.17637298069894314}
|
| 33 |
+
{"time": "13:37:49", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 861.2953844418491, "loss": 0.6097364276647568, "entropy": 0.5986814647912979, "sharp": 0.1646213699132204}
|
| 34 |
+
{"time": "13:39:11", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5808566212654114, "noul_T": 0.841365064765327, "noul_auc": 0.7486455924794518, "noul_acc": 0.696, "noul_f1": 0.7556270096463023, "noul_nll": 0.57935122448951, "noul_ece": 0.061575506865978256, "noul_brier": 0.1974159883600279, "noul_sharp": 0.17260916531085968, "entropy": 0.6056253910064697}
|
| 35 |
+
{"time": "13:42:32", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 851.2136476821952, "loss": 0.5866231825202703, "entropy": 0.6040443293750286, "sharp": 0.1719467956572771}
|
| 36 |
+
{"time": "13:45:47", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 851.9570068978039, "loss": 0.6325390785932541, "entropy": 0.6257338859140873, "sharp": 0.15552806854248047}
|
| 37 |
+
{"time": "13:49:11", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 853.170111477025, "loss": 0.5871488712728024, "entropy": 0.577815443277359, "sharp": 0.19574864581227303}
|
| 38 |
+
{"time": "13:52:36", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 854.0772168467298, "loss": 0.6290334947407246, "entropy": 0.6113212462514639, "sharp": 0.16939380206167698}
|
| 39 |
+
{"time": "13:55:54", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 854.8567721598081, "loss": 0.6088964641094208, "entropy": 0.6100410968065262, "sharp": 0.1543452851474285}
|
| 40 |
+
{"time": "13:59:17", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 855.8002614257731, "loss": 0.629012793302536, "entropy": 0.600356936454773, "sharp": 0.1833872264251113}
|
| 41 |
+
{"time": "14:02:38", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 856.8185098450124, "loss": 0.60341892298311, "entropy": 0.5919076539576054, "sharp": 0.18425844050943851}
|
| 42 |
+
{"time": "14:06:11", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 857.5693821535474, "loss": 0.6246488206088543, "entropy": 0.6165679730474949, "sharp": 0.16228204779326916}
|
| 43 |
+
{"time": "14:09:43", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 858.5631900520607, "loss": 0.6244985517114401, "entropy": 0.6117139812558889, "sharp": 0.16274801176041365}
|
| 44 |
+
{"time": "14:13:02", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 859.0192791997956, "loss": 0.6101237125694752, "entropy": 0.6056259348988533, "sharp": 0.1641486482694745}
|
| 45 |
+
{"time": "14:14:24", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5776541233062744, "noul_T": 0.9134518168237279, "noul_auc": 0.7482574133769894, "noul_acc": 0.704, "noul_f1": 0.765079365079365, "noul_nll": 0.5763804786726833, "noul_ece": 0.04531484305858612, "noul_brier": 0.19632492698913945, "noul_sharp": 0.1842217743396759, "entropy": 0.5919598340988159}
|
| 46 |
+
{"time": "14:17:47", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 851.4969077534847, "loss": 0.6143350675702095, "entropy": 0.6133048012852669, "sharp": 0.16104049514979124}
|
| 47 |
+
{"time": "14:21:02", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 852.1057644594101, "loss": 0.6319963606074452, "entropy": 0.6179205933585763, "sharp": 0.1569443354383111}
|
| 48 |
+
{"time": "14:24:32", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 852.8281467680285, "loss": 0.6231420477852225, "entropy": 0.6066815573722124, "sharp": 0.17096013855189085}
|
| 49 |
+
{"time": "14:27:52", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 853.4843664110587, "loss": 0.6180014163255692, "entropy": 0.6100546419620514, "sharp": 0.16114776860922575}
|
| 50 |
+
{"time": "14:31:27", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 854.0772611668422, "loss": 0.586000457406044, "entropy": 0.6117531713098288, "sharp": 0.17293117381632328}
|
| 51 |
+
{"time": "14:34:43", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 854.607106296014, "loss": 0.6012379117310047, "entropy": 0.5999050606042147, "sharp": 0.18459924310445786}
|
| 52 |
+
{"time": "14:38:02", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 855.2108656691644, "loss": 0.6424654796719551, "entropy": 0.6158636510372162, "sharp": 0.16354532539844513}
|
| 53 |
+
{"time": "14:41:35", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 855.8957361953685, "loss": 0.6105100437998772, "entropy": 0.5965423621237278, "sharp": 0.19001847133040428}
|
| 54 |
+
{"time": "14:45:06", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 856.5078004947596, "loss": 0.5758148171007633, "entropy": 0.5900470390915871, "sharp": 0.1958698369562626}
|
| 55 |
+
{"time": "14:48:39", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 857.1106533423929, "loss": 0.5675816535949707, "entropy": 0.5668441653251648, "sharp": 0.2250992637127638}
|
| 56 |
+
{"time": "14:50:01", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5789852738380432, "noul_T": 0.8641835134060968, "noul_auc": 0.7502826956507063, "noul_acc": 0.702, "noul_f1": 0.7600644122383253, "noul_nll": 0.5770031152516604, "noul_ece": 0.05571204042434693, "noul_brier": 0.19641170153876195, "noul_sharp": 0.17702382802963257, "entropy": 0.6007232069969177}
|
| 57 |
+
{"time": "14:55:31", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 836.8005348893143, "rows_per_s": 1.5464522468215414, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5651941895484924, "noul_T": 0.8614883118654615, "choice_ce": 0.6922976970672607, "choice_T": 1.0608593759639455, "score_ce": 1.0891292095184326, "score_T": 1.0216659864400268, "noul_auc": 0.7393109333333333, "noul_acc": 0.695, "noul_f1": 0.7648419429452583, "noul_nll": 0.5775874657034874, "noul_ece": 0.05573579064011575, "noul_brier": 0.19712157424601043, "noul_sharp": 0.17942030727863312, "entropy": 0.5986356139183044}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"in_proj_z",
|
| 39 |
+
"v_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"down_proj",
|
| 42 |
+
"q_proj",
|
| 43 |
+
"gate_proj",
|
| 44 |
+
"out_proj",
|
| 45 |
+
"up_proj",
|
| 46 |
+
"in_proj_qkv",
|
| 47 |
+
"o_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0003378785331733525,
|
| 4 |
+
-0.0003381400019861758,
|
| 5 |
+
-0.0026545433793216944,
|
| 6 |
+
-0.0017254167469218373,
|
| 7 |
+
-0.0006690291920676827,
|
| 8 |
+
0.0020235059782862663,
|
| 9 |
+
0.001035472727380693,
|
| 10 |
+
-0.0024520151782780886,
|
| 11 |
+
-0.00022121216170489788,
|
| 12 |
+
-0.00040793779771775007,
|
| 13 |
+
3.9059446862665936e-05,
|
| 14 |
+
0.0005792055162601173,
|
| 15 |
+
0.0007219534600153565,
|
| 16 |
+
-0.003824315033853054,
|
| 17 |
+
-0.004488784354180098,
|
| 18 |
+
-0.0051357438787817955,
|
| 19 |
+
-0.0059356484562158585,
|
| 20 |
+
-0.004472075030207634,
|
| 21 |
+
-0.00498128542676568,
|
| 22 |
+
-0.005114296451210976,
|
| 23 |
+
-0.004075711593031883,
|
| 24 |
+
-0.0035568857565522194,
|
| 25 |
+
-0.005701664835214615,
|
| 26 |
+
-0.003578176023438573
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8536592393696032,
|
| 5 |
+
"choice": 1.0600599596812863,
|
| 6 |
+
"score": 0.9467571672038931
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8536592393696032,
|
| 12 |
+
"ce_before": 0.5546435117721558,
|
| 13 |
+
"ce_after": 0.5526026487350464
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 1.0600599596812863,
|
| 18 |
+
"ce_before": 0.6774709224700928,
|
| 19 |
+
"ce_after": 0.6767087578773499
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 0.9467571672038931,
|
| 24 |
+
"ce_before": 1.0601274967193604,
|
| 25 |
+
"ce_after": 1.0592390298843384
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-G-d512-n16k-a075-syn/ckpt-16000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/ckpt-16000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-G-d512-n16k-a075-syn/ckpt-16000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e18/G-d512-n16k-a075",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "G-d512-n16k-a075",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "4c30bbac9daacf951bdd23b1ad7f2ba4c76bcc6496314177f45a0059009778b8",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
|
| 128 |
+
"total_rows": 16000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-16000",
|
| 137 |
+
"rows_seen": 16000,
|
| 138 |
+
"replay_seen": 842,
|
| 139 |
+
"step": 500,
|
| 140 |
+
"tokens": 9113372,
|
| 141 |
+
"tok_per_s": 838.6071492220033,
|
| 142 |
+
"rows_per_s": 1.5497909668527459,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5546435117721558,
|
| 147 |
+
"noul_T": 0.8536592393696032,
|
| 148 |
+
"choice_ce": 0.6774709224700928,
|
| 149 |
+
"choice_T": 1.0600599596812863,
|
| 150 |
+
"score_ce": 1.0601274967193604,
|
| 151 |
+
"score_T": 0.9467571672038931,
|
| 152 |
+
"noul_auc": 0.7474303999999999,
|
| 153 |
+
"noul_acc": 0.706,
|
| 154 |
+
"noul_f1": 0.7772727272727272,
|
| 155 |
+
"noul_nll": 0.5676313229501248,
|
| 156 |
+
"noul_ece": 0.037996120408177375,
|
| 157 |
+
"noul_brier": 0.19365269594103807,
|
| 158 |
+
"noul_sharp": 0.18452808260917664,
|
| 159 |
+
"entropy": 0.5889261960983276
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-a075-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "17:04:38", "event": "start", "data": "data/e18/G-d512-n16k-a075", "data_manifest": {"arm": "G-d512-n16k-a075", "name": null, "train_sha256": "4c30bbac9daacf951bdd23b1ad7f2ba4c76bcc6496314177f45a0059009778b8", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 119.97121095657349}
|
| 2 |
+
{"time": "17:08:05", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 872.0322708261077, "loss": 0.5938443019986153, "entropy": 0.550717493519187, "sharp": 0.23847326636314392}
|
| 3 |
+
{"time": "17:11:32", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 875.4057435393112, "loss": 0.6234635878354311, "entropy": 0.5598232597112656, "sharp": 0.22327384166419506}
|
| 4 |
+
{"time": "17:15:00", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 877.5363235633822, "loss": 0.6085930354893208, "entropy": 0.5877436064183712, "sharp": 0.1926639387384057}
|
| 5 |
+
{"time": "17:18:23", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 879.1642815672654, "loss": 0.576011098921299, "entropy": 0.6400140151381493, "sharp": 0.13992295321077108}
|
| 6 |
+
{"time": "17:21:52", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 881.2546785144235, "loss": 0.5411809347569942, "entropy": 0.5809001699090004, "sharp": 0.2002868503332138}
|
| 7 |
+
{"time": "17:25:13", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 883.0682555497644, "loss": 0.5916874818503857, "entropy": 0.6490094624459743, "sharp": 0.13038080837577581}
|
| 8 |
+
{"time": "17:28:42", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 882.9978743867365, "loss": 0.5733799301087856, "entropy": 0.5609774179756641, "sharp": 0.2185530737042427}
|
| 9 |
+
{"time": "17:32:07", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 883.1005839135017, "loss": 0.6565701058134437, "entropy": 0.6008156426250935, "sharp": 0.18316218070685863}
|
| 10 |
+
{"time": "17:35:37", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 884.4759886502945, "loss": 0.6326943729072809, "entropy": 0.5657059103250504, "sharp": 0.2127220332622528}
|
| 11 |
+
{"time": "17:39:05", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 884.236438931958, "loss": 0.5607275497168303, "entropy": 0.6241490747779608, "sharp": 0.15848525427281857}
|
| 12 |
+
{"time": "17:40:28", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5817887187004089, "noul_T": 0.7974794803380608, "noul_auc": 0.7414895951123188, "noul_acc": 0.696, "noul_f1": 0.7617554858934169, "noul_nll": 0.5862313076108694, "noul_ece": 0.07873240214586258, "noul_brier": 0.20002084117316118, "noul_sharp": 0.16990309953689575, "entropy": 0.613395631313324}
|
| 13 |
+
{"time": "17:43:51", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 853.9506085822945, "loss": 0.6271498166024685, "entropy": 0.5876873098313808, "sharp": 0.1962280748412013}
|
| 14 |
+
{"time": "17:47:27", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 857.402840728958, "loss": 0.5846296865493059, "entropy": 0.5995341166853905, "sharp": 0.1763457152992487}
|
| 15 |
+
{"time": "17:51:01", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 860.4980448566976, "loss": 0.5598829463124275, "entropy": 0.5502076372504234, "sharp": 0.21965670958161354}
|
| 16 |
+
{"time": "17:54:28", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 862.0586210620685, "loss": 0.6096249129623175, "entropy": 0.6195895150303841, "sharp": 0.1619223952293396}
|
| 17 |
+
{"time": "17:57:58", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 863.5844440640948, "loss": 0.6493042409420013, "entropy": 0.6322226636111736, "sharp": 0.14232324995100498}
|
| 18 |
+
{"time": "18:01:19", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 864.962024010448, "loss": 0.6248074807226658, "entropy": 0.5708954501897097, "sharp": 0.20338837802410126}
|
| 19 |
+
{"time": "18:04:48", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 866.3409173629072, "loss": 0.6037216018885374, "entropy": 0.5924348682165146, "sharp": 0.18726175650954247}
|
| 20 |
+
{"time": "18:08:13", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 867.6657085319231, "loss": 0.5961871184408665, "entropy": 0.5825368985533714, "sharp": 0.19010062143206596}
|
| 21 |
+
{"time": "18:11:39", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 868.427505713956, "loss": 0.6275307163596153, "entropy": 0.5924929901957512, "sharp": 0.17990336753427982}
|
| 22 |
+
{"time": "18:15:06", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 869.4509848621333, "loss": 0.5739761553704739, "entropy": 0.5982164461165667, "sharp": 0.19135244749486446}
|
| 23 |
+
{"time": "18:16:28", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5715430974960327, "noul_T": 0.8199112685604122, "noul_auc": 0.7479873757404939, "noul_acc": 0.7, "noul_f1": 0.7713414634146342, "noul_nll": 0.5747418907731772, "noul_ece": 0.04500952571630479, "noul_brier": 0.19593098972319664, "noul_sharp": 0.17675723135471344, "entropy": 0.6011351943016052}
|
| 24 |
+
{"time": "18:19:59", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 854.6451645404328, "loss": 0.5610256269574165, "entropy": 0.5567095745354891, "sharp": 0.2205702606588602}
|
| 25 |
+
{"time": "18:23:21", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 855.8578892470092, "loss": 0.603161983191967, "entropy": 0.5846798419952393, "sharp": 0.19453026913106441}
|
| 26 |
+
{"time": "18:26:40", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 857.098237579371, "loss": 0.5733331013470888, "entropy": 0.565919129177928, "sharp": 0.22536353394389153}
|
| 27 |
+
{"time": "18:30:04", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 858.2433698346404, "loss": 0.574288371950388, "entropy": 0.5778393354266882, "sharp": 0.20217853598296642}
|
| 28 |
+
{"time": "18:33:43", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 859.1371914815197, "loss": 0.6029765270650387, "entropy": 0.6025132574141026, "sharp": 0.1749593149870634}
|
| 29 |
+
{"time": "18:36:56", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 859.9896049247614, "loss": 0.5813244134187698, "entropy": 0.6036417782306671, "sharp": 0.16531551722437143}
|
| 30 |
+
{"time": "18:40:22", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 860.8627140262131, "loss": 0.6397402584552765, "entropy": 0.6044099386781454, "sharp": 0.16990405786782503}
|
| 31 |
+
{"time": "18:43:45", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 861.8275306776967, "loss": 0.6233418276533484, "entropy": 0.5752152260392904, "sharp": 0.20379032380878925}
|
| 32 |
+
{"time": "18:47:15", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 862.7841110651692, "loss": 0.6249094009399414, "entropy": 0.5605471394956112, "sharp": 0.21292312443256378}
|
| 33 |
+
{"time": "18:50:43", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 863.7524442202757, "loss": 0.5684554595500231, "entropy": 0.5949028339236975, "sharp": 0.15776244830340147}
|
| 34 |
+
{"time": "18:52:05", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5750628709793091, "noul_T": 0.7366794628618392, "noul_auc": 0.7480380077973368, "noul_acc": 0.694, "noul_f1": 0.7544141252006421, "noul_nll": 0.5781629832684994, "noul_ece": 0.07445384293794634, "noul_brier": 0.1976498937080665, "noul_sharp": 0.15978877246379852, "entropy": 0.61398845911026}
|
| 35 |
+
{"time": "18:55:25", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 853.5543211574739, "loss": 0.5711853839457035, "entropy": 0.5751893930137157, "sharp": 0.1900377068668604}
|
| 36 |
+
{"time": "18:58:40", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 854.263741666618, "loss": 0.6429812433198094, "entropy": 0.6241484507918358, "sharp": 0.1509555783122778}
|
| 37 |
+
{"time": "19:02:05", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 855.4066012289314, "loss": 0.576201967895031, "entropy": 0.568741649389267, "sharp": 0.21256351843476295}
|
| 38 |
+
{"time": "19:05:30", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 856.2615083354541, "loss": 0.6337269730865955, "entropy": 0.597293846309185, "sharp": 0.1770805399864912}
|
| 39 |
+
{"time": "19:08:47", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 856.9905595319922, "loss": 0.6125753819942474, "entropy": 0.5950655601918697, "sharp": 0.17312759812921286}
|
| 40 |
+
{"time": "19:12:11", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 857.8731407785207, "loss": 0.6160881221294403, "entropy": 0.6157016046345234, "sharp": 0.16258691810071468}
|
| 41 |
+
{"time": "19:15:32", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 858.8493495055507, "loss": 0.6139410762116313, "entropy": 0.5616400893777609, "sharp": 0.21455884911119938}
|
| 42 |
+
{"time": "19:19:04", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 859.5760263496621, "loss": 0.6049833595752716, "entropy": 0.5889336355030537, "sharp": 0.1852509193122387}
|
| 43 |
+
{"time": "19:22:36", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 860.5351731993973, "loss": 0.630169190466404, "entropy": 0.6101519577205181, "sharp": 0.15933663118630648}
|
| 44 |
+
{"time": "19:25:54", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 860.9799019992898, "loss": 0.5913512520492077, "entropy": 0.5976393967866898, "sharp": 0.1696687750518322}
|
| 45 |
+
{"time": "19:27:16", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5675866007804871, "noul_T": 0.9069650430496604, "noul_auc": 0.7503839597643921, "noul_acc": 0.706, "noul_f1": 0.7769347496206374, "noul_nll": 0.5705840392410755, "noul_ece": 0.05276812362670901, "noul_brier": 0.19499460868171228, "noul_sharp": 0.1883695125579834, "entropy": 0.5830758213996887}
|
| 46 |
+
{"time": "19:30:39", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 853.4116988084041, "loss": 0.602540947496891, "entropy": 0.6002940088510513, "sharp": 0.17506624478846788}
|
| 47 |
+
{"time": "19:33:54", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 854.0355227702181, "loss": 0.6473210928961635, "entropy": 0.6179999969899654, "sharp": 0.1565154790878296}
|
| 48 |
+
{"time": "19:37:23", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 854.7727960339807, "loss": 0.5979703152552247, "entropy": 0.5897938935086131, "sharp": 0.17914921045303345}
|
| 49 |
+
{"time": "19:40:43", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 855.4342783052027, "loss": 0.6045144088566303, "entropy": 0.5992164053022861, "sharp": 0.1674152882769704}
|
| 50 |
+
{"time": "19:44:17", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 856.0037700341809, "loss": 0.5585590526461601, "entropy": 0.5925435498356819, "sharp": 0.1894065123051405}
|
| 51 |
+
{"time": "19:47:34", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 856.4955677056948, "loss": 0.5895522497594357, "entropy": 0.5956769846379757, "sharp": 0.1877808328717947}
|
| 52 |
+
{"time": "19:50:52", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 857.0722376578864, "loss": 0.6110889986157417, "entropy": 0.5910369232296944, "sharp": 0.1948243472725153}
|
| 53 |
+
{"time": "19:54:25", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 857.7406191565195, "loss": 0.5944768842309713, "entropy": 0.5751582439988852, "sharp": 0.20663999672979116}
|
| 54 |
+
{"time": "19:57:55", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 858.3614960688146, "loss": 0.5948281362652779, "entropy": 0.5811646357178688, "sharp": 0.19812480174005032}
|
| 55 |
+
{"time": "20:01:28", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 858.9836778867082, "loss": 0.5572592057287693, "entropy": 0.5433378294110298, "sharp": 0.24021293222904205}
|
| 56 |
+
{"time": "20:02:50", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5689151883125305, "noul_T": 0.870908825553566, "noul_auc": 0.7529155626065384, "noul_acc": 0.708, "noul_f1": 0.7697160883280757, "noul_nll": 0.5696612951830029, "noul_ece": 0.03668692427873612, "noul_brier": 0.19442570594667258, "noul_sharp": 0.1826082020998001, "entropy": 0.5899560451507568}
|
| 57 |
+
{"time": "20:05:53", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 838.6071492220033, "rows_per_s": 1.5497909668527459, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5546435117721558, "noul_T": 0.8536592393696032, "choice_ce": 0.6774709224700928, "choice_T": 1.0600599596812863, "score_ce": 1.0601274967193604, "score_T": 0.9467571672038931, "noul_auc": 0.7474303999999999, "noul_acc": 0.706, "noul_f1": 0.7772727272727272, "noul_nll": 0.5676313229501248, "noul_ece": 0.037996120408177375, "noul_brier": 0.19365269594103807, "noul_sharp": 0.18452808260917664, "entropy": 0.5889261960983276}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"gate_proj",
|
| 39 |
+
"down_proj",
|
| 40 |
+
"up_proj",
|
| 41 |
+
"in_proj_qkv",
|
| 42 |
+
"in_proj_z",
|
| 43 |
+
"out_proj",
|
| 44 |
+
"q_proj",
|
| 45 |
+
"k_proj",
|
| 46 |
+
"o_proj",
|
| 47 |
+
"v_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0003777900419663638,
|
| 4 |
+
-0.0003780527913477272,
|
| 5 |
+
-0.0028317768592387438,
|
| 6 |
+
-0.0021922902669757605,
|
| 7 |
+
-0.0007482326473109424,
|
| 8 |
+
0.001780627528205514,
|
| 9 |
+
0.001410171971656382,
|
| 10 |
+
-0.0030618044547736645,
|
| 11 |
+
-0.0004620991530828178,
|
| 12 |
+
-0.0003244285471737385,
|
| 13 |
+
9.252070594811812e-05,
|
| 14 |
+
0.0005214349366724491,
|
| 15 |
+
-7.778857980156317e-05,
|
| 16 |
+
-0.0037078019231557846,
|
| 17 |
+
-0.004459568299353123,
|
| 18 |
+
-0.005217264872044325,
|
| 19 |
+
-0.005987458396703005,
|
| 20 |
+
-0.004444331396371126,
|
| 21 |
+
-0.004958426114171743,
|
| 22 |
+
-0.005050053354352713,
|
| 23 |
+
-0.003836591960862279,
|
| 24 |
+
-0.0034511960111558437,
|
| 25 |
+
-0.00576677406206727,
|
| 26 |
+
-0.0035300070885568857
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8425032178558856,
|
| 5 |
+
"choice": 1.0807915279439129,
|
| 6 |
+
"score": 0.971387470364942
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8425032178558856,
|
| 12 |
+
"ce_before": 0.5593699216842651,
|
| 13 |
+
"ce_after": 0.5569900870323181
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 1.0807915279439129,
|
| 18 |
+
"ce_before": 0.6868921518325806,
|
| 19 |
+
"ce_after": 0.6854957342147827
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 0.971387470364942,
|
| 24 |
+
"ce_before": 1.064429521560669,
|
| 25 |
+
"ce_after": 1.0641733407974243
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-G-d512-n16k-syn/ckpt-16000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/ckpt-16000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-G-d512-n16k-syn/ckpt-16000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e18/G-d512-n16k",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "G-d512-n16k",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "ef94315960694ce58bdc6d61b8665b7bf46f4e2996b9e82503e1d17bb87a14b7",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
|
| 128 |
+
"total_rows": 16000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-16000",
|
| 137 |
+
"rows_seen": 16000,
|
| 138 |
+
"replay_seen": 842,
|
| 139 |
+
"step": 500,
|
| 140 |
+
"tokens": 9113372,
|
| 141 |
+
"tok_per_s": 834.3555531518748,
|
| 142 |
+
"rows_per_s": 1.5419337895036624,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5593699216842651,
|
| 147 |
+
"noul_T": 0.8425032178558856,
|
| 148 |
+
"choice_ce": 0.6868921518325806,
|
| 149 |
+
"choice_T": 1.0807915279439129,
|
| 150 |
+
"score_ce": 1.064429521560669,
|
| 151 |
+
"score_T": 0.971387470364942,
|
| 152 |
+
"noul_auc": 0.7457685333333333,
|
| 153 |
+
"noul_acc": 0.697,
|
| 154 |
+
"noul_f1": 0.7681713848508034,
|
| 155 |
+
"noul_nll": 0.5711014421954751,
|
| 156 |
+
"noul_ece": 0.04758954861760138,
|
| 157 |
+
"noul_brier": 0.19464147341873098,
|
| 158 |
+
"noul_sharp": 0.17954327166080475,
|
| 159 |
+
"entropy": 0.5962615609169006
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-G-d512-n16k-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "01:17:49", "event": "start", "data": "data/e18/G-d512-n16k", "data_manifest": {"arm": "G-d512-n16k", "name": null, "train_sha256": "ef94315960694ce58bdc6d61b8665b7bf46f4e2996b9e82503e1d17bb87a14b7", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 111.50794792175293}
|
| 2 |
+
{"time": "01:21:16", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 868.959319411254, "loss": 0.6013306770473719, "entropy": 0.586697407066822, "sharp": 0.20414860174059868}
|
| 3 |
+
{"time": "01:24:45", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 870.4163929812999, "loss": 0.6118059046566486, "entropy": 0.5823066495358944, "sharp": 0.21134249307215214}
|
| 4 |
+
{"time": "01:28:14", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 872.5182943942087, "loss": 0.6056998502463102, "entropy": 0.5301483161747456, "sharp": 0.24638377223163843}
|
| 5 |
+
{"time": "01:31:38", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 874.4766136242328, "loss": 0.5798004642128944, "entropy": 0.6060839369893074, "sharp": 0.1760128540918231}
|
| 6 |
+
{"time": "01:35:08", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 876.8455535138042, "loss": 0.5605190563946962, "entropy": 0.5915488190948963, "sharp": 0.19267505407333374}
|
| 7 |
+
{"time": "01:38:30", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 878.914844472687, "loss": 0.5933446958661079, "entropy": 0.6282293498516083, "sharp": 0.1574245449155569}
|
| 8 |
+
{"time": "01:41:58", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 879.3909392357791, "loss": 0.5827849097549915, "entropy": 0.5883386991918087, "sharp": 0.19046484306454659}
|
| 9 |
+
{"time": "01:45:24", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 879.303730304069, "loss": 0.644519405439496, "entropy": 0.5615602526813745, "sharp": 0.21720010787248611}
|
| 10 |
+
{"time": "01:48:57", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 880.4313526854997, "loss": 0.6346069760620594, "entropy": 0.6044753231108189, "sharp": 0.16464239731431007}
|
| 11 |
+
{"time": "01:52:26", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 879.7027959637883, "loss": 0.5707364976406097, "entropy": 0.6028544828295708, "sharp": 0.17858518660068512}
|
| 12 |
+
{"time": "01:53:49", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5834838151931763, "noul_T": 0.9739584607326618, "noul_auc": 0.7401056522252789, "noul_acc": 0.702, "noul_f1": 0.7553366174055829, "noul_nll": 0.5868862022981047, "noul_ece": 0.06867078310251236, "noul_brier": 0.201077582959189, "noul_sharp": 0.18924878537654877, "entropy": 0.5878397822380066}
|
| 13 |
+
{"time": "01:57:13", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 849.6602179057368, "loss": 0.6122045703232288, "entropy": 0.6132514104247093, "sharp": 0.16648668237030506}
|
| 14 |
+
{"time": "02:00:49", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 853.3311932938715, "loss": 0.5809519980102777, "entropy": 0.5973322652280331, "sharp": 0.17421594634652138}
|
| 15 |
+
{"time": "02:04:25", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 856.4749768352304, "loss": 0.5661034509539604, "entropy": 0.5788102969527245, "sharp": 0.19747170247137547}
|
| 16 |
+
{"time": "02:07:53", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 857.9706092296371, "loss": 0.6186694260686636, "entropy": 0.6239436753094196, "sharp": 0.15775544196367264}
|
| 17 |
+
{"time": "02:11:24", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 859.4055571093138, "loss": 0.6403868682682514, "entropy": 0.6394223645329475, "sharp": 0.13979184348136187}
|
| 18 |
+
{"time": "02:14:46", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 860.6902517402559, "loss": 0.6256495323032141, "entropy": 0.5635669194161892, "sharp": 0.20874473080039024}
|
| 19 |
+
{"time": "02:18:16", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 862.0412909061776, "loss": 0.6126365400850773, "entropy": 0.6186674982309341, "sharp": 0.15412371326237917}
|
| 20 |
+
{"time": "02:21:44", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 863.1854001037167, "loss": 0.5879520401358604, "entropy": 0.5987095274031162, "sharp": 0.17986678704619408}
|
| 21 |
+
{"time": "02:25:12", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 863.750303395381, "loss": 0.6287038549780846, "entropy": 0.6031125374138355, "sharp": 0.17224498558789492}
|
| 22 |
+
{"time": "02:28:40", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 864.6322404337253, "loss": 0.5912205968052149, "entropy": 0.6146058402955532, "sharp": 0.17249909974634647}
|
| 23 |
+
{"time": "02:30:03", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5751998424530029, "noul_T": 0.8141045240377831, "noul_auc": 0.7468397157853875, "noul_acc": 0.706, "noul_f1": 0.7699530516431925, "noul_nll": 0.5782119868248701, "noul_ece": 0.05508123040199281, "noul_brier": 0.19686475400000494, "noul_sharp": 0.1735110580921173, "entropy": 0.6052521467208862}
|
| 24 |
+
{"time": "02:33:35", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 849.8384286182772, "loss": 0.5968205723911524, "entropy": 0.5830590762197971, "sharp": 0.1971623096615076}
|
| 25 |
+
{"time": "02:36:59", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 850.9387536298985, "loss": 0.606288019567728, "entropy": 0.5797390006482601, "sharp": 0.19601893704384565}
|
| 26 |
+
{"time": "02:40:19", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 852.1352041246533, "loss": 0.5891094021499157, "entropy": 0.5848966054618359, "sharp": 0.20710623264312744}
|
| 27 |
+
{"time": "02:43:44", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 853.2044064206584, "loss": 0.5891074351966381, "entropy": 0.5942482110112906, "sharp": 0.1876743659377098}
|
| 28 |
+
{"time": "02:47:25", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 854.0766021849222, "loss": 0.6180044487118721, "entropy": 0.6009003445506096, "sharp": 0.17110509984195232}
|
| 29 |
+
{"time": "02:50:39", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 854.9196349405474, "loss": 0.5874475985765457, "entropy": 0.6006395071744919, "sharp": 0.17370467446744442}
|
| 30 |
+
{"time": "02:54:07", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 855.7687336142741, "loss": 0.6360086761415005, "entropy": 0.624266087077558, "sharp": 0.15217309352010489}
|
| 31 |
+
{"time": "02:57:31", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 856.7515276080856, "loss": 0.6349860578775406, "entropy": 0.6078876182436943, "sharp": 0.17538239248096943}
|
| 32 |
+
{"time": "03:01:02", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 857.6715277053227, "loss": 0.6174807772040367, "entropy": 0.5940439812839031, "sharp": 0.17607913631945848}
|
| 33 |
+
{"time": "03:04:30", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 858.7077063212081, "loss": 0.5889724995940924, "entropy": 0.5827264972031116, "sharp": 0.16978934034705162}
|
| 34 |
+
{"time": "03:05:53", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5708449482917786, "noul_T": 0.8591901324922298, "noul_auc": 0.7487299792408567, "noul_acc": 0.698, "noul_f1": 0.7636932707355243, "noul_nll": 0.5731498726606369, "noul_ece": 0.0521172947883606, "noul_brier": 0.19556991272027313, "noul_sharp": 0.179712176322937, "entropy": 0.5937709808349609}
|
| 35 |
+
{"time": "03:09:14", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 848.663010651659, "loss": 0.57573894970119, "entropy": 0.6039266008883715, "sharp": 0.16989239118993282}
|
| 36 |
+
{"time": "03:12:30", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 849.4392086439179, "loss": 0.6336011225357652, "entropy": 0.6147672161459923, "sharp": 0.16693622432649136}
|
| 37 |
+
{"time": "03:15:55", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 850.6026163768321, "loss": 0.5828944221138954, "entropy": 0.5823952555656433, "sharp": 0.18948199227452278}
|
| 38 |
+
{"time": "03:19:21", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 851.4844653116731, "loss": 0.6303686816245317, "entropy": 0.6011038944125175, "sharp": 0.17796625010669231}
|
| 39 |
+
{"time": "03:22:39", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 852.2760947389676, "loss": 0.610909715294838, "entropy": 0.6145693995058537, "sharp": 0.1515007670968771}
|
| 40 |
+
{"time": "03:26:03", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 853.2387589736784, "loss": 0.6218335591256618, "entropy": 0.5991636477410793, "sharp": 0.18547695875167847}
|
| 41 |
+
{"time": "03:29:24", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 854.2662953773039, "loss": 0.6039894968271255, "entropy": 0.5765786161646247, "sharp": 0.20314838737249374}
|
| 42 |
+
{"time": "03:32:57", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 855.0354649663262, "loss": 0.615525808185339, "entropy": 0.6060565896332264, "sharp": 0.1717838179320097}
|
| 43 |
+
{"time": "03:36:30", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 856.0518368629158, "loss": 0.6292471252381802, "entropy": 0.6094392910599709, "sharp": 0.1642614183947444}
|
| 44 |
+
{"time": "03:39:49", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 856.5223906346504, "loss": 0.6033096313476562, "entropy": 0.6068289577960968, "sharp": 0.162236537784338}
|
| 45 |
+
{"time": "03:41:11", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5712239742279053, "noul_T": 0.8859973212858163, "noul_auc": 0.7484093095475184, "noul_acc": 0.704, "noul_f1": 0.7708978328173375, "noul_nll": 0.5736466155424714, "noul_ece": 0.04043909198045731, "noul_brier": 0.19564575867200296, "noul_sharp": 0.18395410478115082, "entropy": 0.5902425646781921}
|
| 46 |
+
{"time": "03:44:35", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 849.0388859689781, "loss": 0.608968123793602, "entropy": 0.6042648628354073, "sharp": 0.1703565027564764}
|
| 47 |
+
{"time": "03:47:50", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 849.6772174297872, "loss": 0.6396214291453362, "entropy": 0.6148591954261065, "sharp": 0.160504967905581}
|
| 48 |
+
{"time": "03:51:21", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 850.3753067297807, "loss": 0.6124422987923026, "entropy": 0.5988048352301121, "sharp": 0.1735254293307662}
|
| 49 |
+
{"time": "03:54:42", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 851.0258856493341, "loss": 0.6124591641128063, "entropy": 0.6065691262483597, "sharp": 0.16226070281118155}
|
| 50 |
+
{"time": "03:58:17", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 851.5854371946922, "loss": 0.575305238366127, "entropy": 0.6048825699836016, "sharp": 0.17797653563320637}
|
| 51 |
+
{"time": "04:01:35", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 852.0878215105005, "loss": 0.5959524214267731, "entropy": 0.5974207315593958, "sharp": 0.18652764149010181}
|
| 52 |
+
{"time": "04:04:54", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 852.656360844423, "loss": 0.6249396838247776, "entropy": 0.6030560433864594, "sharp": 0.18152933940291405}
|
| 53 |
+
{"time": "04:08:28", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 853.3290721671551, "loss": 0.6032074559479952, "entropy": 0.590003527700901, "sharp": 0.1944424668326974}
|
| 54 |
+
{"time": "04:11:59", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 853.9794037505511, "loss": 0.586712434887886, "entropy": 0.5855944007635117, "sharp": 0.1965925581753254}
|
| 55 |
+
{"time": "04:15:33", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 854.5933919788708, "loss": 0.5622827149927616, "entropy": 0.5558024868369102, "sharp": 0.23326499201357365}
|
| 56 |
+
{"time": "04:16:56", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5732556581497192, "noul_T": 0.8465915574517751, "noul_auc": 0.7515653744240605, "noul_acc": 0.704, "noul_f1": 0.7635782747603834, "noul_nll": 0.5737326111495494, "noul_ece": 0.05849021238088608, "noul_brier": 0.19552273307461854, "noul_sharp": 0.17710581421852112, "entropy": 0.5980157852172852}
|
| 57 |
+
{"time": "04:19:56", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 834.3555531518748, "rows_per_s": 1.5419337895036624, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5593699216842651, "noul_T": 0.8425032178558856, "choice_ce": 0.6868921518325806, "choice_T": 1.0807915279439129, "score_ce": 1.064429521560669, "score_T": 0.971387470364942, "noul_auc": 0.7457685333333333, "noul_acc": 0.697, "noul_f1": 0.7681713848508034, "noul_nll": 0.5711014421954751, "noul_ece": 0.04758954861760138, "noul_brier": 0.19464147341873098, "noul_sharp": 0.17954327166080475, "entropy": 0.5962615609169006}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"out_proj",
|
| 39 |
+
"k_proj",
|
| 40 |
+
"up_proj",
|
| 41 |
+
"gate_proj",
|
| 42 |
+
"in_proj_qkv",
|
| 43 |
+
"q_proj",
|
| 44 |
+
"in_proj_z",
|
| 45 |
+
"o_proj",
|
| 46 |
+
"v_proj",
|
| 47 |
+
"down_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/adapter_vllm/decision_head.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": [
|
| 3 |
+
0.0004198715032543987,
|
| 4 |
+
-0.000420135329477489,
|
| 5 |
+
-0.002468167105689645,
|
| 6 |
+
-0.0016938160406425595,
|
| 7 |
+
-0.00044213730143383145,
|
| 8 |
+
0.002100521931424737,
|
| 9 |
+
0.0015469910576939583,
|
| 10 |
+
-0.0022080650087445974,
|
| 11 |
+
-0.0006371029303409159,
|
| 12 |
+
-0.0005100148264318705,
|
| 13 |
+
-0.00018270142027176917,
|
| 14 |
+
0.0004760129959322512,
|
| 15 |
+
-0.00021764486155007035,
|
| 16 |
+
-0.0038659980054944754,
|
| 17 |
+
-0.0044587142765522,
|
| 18 |
+
-0.005133585538715124,
|
| 19 |
+
-0.005951011553406715,
|
| 20 |
+
-0.004421474877744913,
|
| 21 |
+
-0.004959343001246452,
|
| 22 |
+
-0.005074902903288603,
|
| 23 |
+
-0.003985952120274305,
|
| 24 |
+
-0.0034883671905845404,
|
| 25 |
+
-0.00571385957300663,
|
| 26 |
+
-0.0035285777412354946
|
| 27 |
+
],
|
| 28 |
+
"verbalizer_ids": [
|
| 29 |
+
3721,
|
| 30 |
+
1802,
|
| 31 |
+
15,
|
| 32 |
+
16,
|
| 33 |
+
17,
|
| 34 |
+
18,
|
| 35 |
+
19,
|
| 36 |
+
20,
|
| 37 |
+
32,
|
| 38 |
+
33,
|
| 39 |
+
34,
|
| 40 |
+
35,
|
| 41 |
+
36,
|
| 42 |
+
37,
|
| 43 |
+
38,
|
| 44 |
+
39,
|
| 45 |
+
40,
|
| 46 |
+
41,
|
| 47 |
+
42,
|
| 48 |
+
43,
|
| 49 |
+
44,
|
| 50 |
+
45,
|
| 51 |
+
46,
|
| 52 |
+
47
|
| 53 |
+
],
|
| 54 |
+
"slots": {
|
| 55 |
+
"num_slots": 24,
|
| 56 |
+
"ranges": {
|
| 57 |
+
"noul": [
|
| 58 |
+
0,
|
| 59 |
+
2
|
| 60 |
+
],
|
| 61 |
+
"score": [
|
| 62 |
+
2,
|
| 63 |
+
8
|
| 64 |
+
],
|
| 65 |
+
"choice": [
|
| 66 |
+
8,
|
| 67 |
+
24
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"verbalizers": [
|
| 71 |
+
"false",
|
| 72 |
+
"true",
|
| 73 |
+
"0",
|
| 74 |
+
"1",
|
| 75 |
+
"2",
|
| 76 |
+
"3",
|
| 77 |
+
"4",
|
| 78 |
+
"5",
|
| 79 |
+
"A",
|
| 80 |
+
"B",
|
| 81 |
+
"C",
|
| 82 |
+
"D",
|
| 83 |
+
"E",
|
| 84 |
+
"F",
|
| 85 |
+
"G",
|
| 86 |
+
"H",
|
| 87 |
+
"I",
|
| 88 |
+
"J",
|
| 89 |
+
"K",
|
| 90 |
+
"L",
|
| 91 |
+
"M",
|
| 92 |
+
"N",
|
| 93 |
+
"O",
|
| 94 |
+
"P"
|
| 95 |
+
],
|
| 96 |
+
"template_version": "bare-v1"
|
| 97 |
+
},
|
| 98 |
+
"note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
|
| 99 |
+
}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/calibration.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": 1,
|
| 3 |
+
"per_kind": {
|
| 4 |
+
"noul": 0.8479683064850129,
|
| 5 |
+
"choice": 1.0268918846698019,
|
| 6 |
+
"score": 0.980406546626907
|
| 7 |
+
},
|
| 8 |
+
"fit": {
|
| 9 |
+
"noul": {
|
| 10 |
+
"n": 1087,
|
| 11 |
+
"T": 0.8479683064850129,
|
| 12 |
+
"ce_before": 0.5609807372093201,
|
| 13 |
+
"ce_after": 0.558741569519043
|
| 14 |
+
},
|
| 15 |
+
"choice": {
|
| 16 |
+
"n": 60,
|
| 17 |
+
"T": 1.0268918846698019,
|
| 18 |
+
"ce_before": 0.6825830936431885,
|
| 19 |
+
"ce_after": 0.6824300289154053
|
| 20 |
+
},
|
| 21 |
+
"score": {
|
| 22 |
+
"n": 53,
|
| 23 |
+
"T": 0.980406546626907,
|
| 24 |
+
"ce_before": 1.060854196548462,
|
| 25 |
+
"ce_after": 1.0607378482818604
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"source": "models/e16/S1-G-d64-n16k-syn/ckpt-16000"
|
| 29 |
+
}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/ckpt-16000/judge_config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_path": "/root/models/Qwen3.5-9B",
|
| 3 |
+
"hidden_size": 4096,
|
| 4 |
+
"slots": {
|
| 5 |
+
"num_slots": 24,
|
| 6 |
+
"ranges": {
|
| 7 |
+
"noul": [
|
| 8 |
+
0,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"score": [
|
| 12 |
+
2,
|
| 13 |
+
8
|
| 14 |
+
],
|
| 15 |
+
"choice": [
|
| 16 |
+
8,
|
| 17 |
+
24
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"verbalizers": [
|
| 21 |
+
"false",
|
| 22 |
+
"true",
|
| 23 |
+
"0",
|
| 24 |
+
"1",
|
| 25 |
+
"2",
|
| 26 |
+
"3",
|
| 27 |
+
"4",
|
| 28 |
+
"5",
|
| 29 |
+
"A",
|
| 30 |
+
"B",
|
| 31 |
+
"C",
|
| 32 |
+
"D",
|
| 33 |
+
"E",
|
| 34 |
+
"F",
|
| 35 |
+
"G",
|
| 36 |
+
"H",
|
| 37 |
+
"I",
|
| 38 |
+
"J",
|
| 39 |
+
"K",
|
| 40 |
+
"L",
|
| 41 |
+
"M",
|
| 42 |
+
"N",
|
| 43 |
+
"O",
|
| 44 |
+
"P"
|
| 45 |
+
],
|
| 46 |
+
"template_version": "bare-v1"
|
| 47 |
+
},
|
| 48 |
+
"verbalizer_ids": [
|
| 49 |
+
3721,
|
| 50 |
+
1802,
|
| 51 |
+
15,
|
| 52 |
+
16,
|
| 53 |
+
17,
|
| 54 |
+
18,
|
| 55 |
+
19,
|
| 56 |
+
20,
|
| 57 |
+
32,
|
| 58 |
+
33,
|
| 59 |
+
34,
|
| 60 |
+
35,
|
| 61 |
+
36,
|
| 62 |
+
37,
|
| 63 |
+
38,
|
| 64 |
+
39,
|
| 65 |
+
40,
|
| 66 |
+
41,
|
| 67 |
+
42,
|
| 68 |
+
43,
|
| 69 |
+
44,
|
| 70 |
+
45,
|
| 71 |
+
46,
|
| 72 |
+
47
|
| 73 |
+
],
|
| 74 |
+
"kinds": [
|
| 75 |
+
"noul",
|
| 76 |
+
"choice",
|
| 77 |
+
"score"
|
| 78 |
+
],
|
| 79 |
+
"model_name": "jev-judge-qwen35-9b",
|
| 80 |
+
"model_version": "0.8.0",
|
| 81 |
+
"lm_head_included": true,
|
| 82 |
+
"weights_mode": "unmerged",
|
| 83 |
+
"adapter_subfolder": "adapter",
|
| 84 |
+
"lora": {
|
| 85 |
+
"r": 16,
|
| 86 |
+
"lora_alpha": 32,
|
| 87 |
+
"lora_dropout": 0.05,
|
| 88 |
+
"target_modules": [
|
| 89 |
+
"down_proj",
|
| 90 |
+
"gate_proj",
|
| 91 |
+
"in_proj_qkv",
|
| 92 |
+
"in_proj_z",
|
| 93 |
+
"k_proj",
|
| 94 |
+
"o_proj",
|
| 95 |
+
"out_proj",
|
| 96 |
+
"q_proj",
|
| 97 |
+
"up_proj",
|
| 98 |
+
"v_proj"
|
| 99 |
+
]
|
| 100 |
+
},
|
| 101 |
+
"trained_stage": "e16",
|
| 102 |
+
"source_checkpoint": "models/e16/S1-G-d64-n16k-syn/ckpt-16000",
|
| 103 |
+
"softcap": null,
|
| 104 |
+
"e16": {
|
| 105 |
+
"data": "data/e18/G-d64-n16k",
|
| 106 |
+
"data_manifest": {
|
| 107 |
+
"arm": "G-d64-n16k",
|
| 108 |
+
"name": null,
|
| 109 |
+
"train_sha256": "f66315602eb898d5fee95f7929164fe84183d5f3744ac956e10355b96ffee370",
|
| 110 |
+
"n_train": 16000
|
| 111 |
+
},
|
| 112 |
+
"init": "released",
|
| 113 |
+
"objective": "s-log",
|
| 114 |
+
"sigma": 0.5,
|
| 115 |
+
"K": 8,
|
| 116 |
+
"reward": "brier",
|
| 117 |
+
"clip": 0.2,
|
| 118 |
+
"passes": 1,
|
| 119 |
+
"lr": 0.0001,
|
| 120 |
+
"lr_head": 0.0002,
|
| 121 |
+
"rows_per_step": 32,
|
| 122 |
+
"train_seed": 0,
|
| 123 |
+
"replay": "data/e16/replay/train.jsonl",
|
| 124 |
+
"replay_frac": 0.05,
|
| 125 |
+
"replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
|
| 126 |
+
"max_len": 3072,
|
| 127 |
+
"schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
|
| 128 |
+
"total_rows": 16000,
|
| 129 |
+
"lora": {
|
| 130 |
+
"r": 16,
|
| 131 |
+
"lora_alpha": 32,
|
| 132 |
+
"lora_dropout": 0.05
|
| 133 |
+
},
|
| 134 |
+
"trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
|
| 135 |
+
"event": "ckpt",
|
| 136 |
+
"tag": "ckpt-16000",
|
| 137 |
+
"rows_seen": 16000,
|
| 138 |
+
"replay_seen": 842,
|
| 139 |
+
"step": 500,
|
| 140 |
+
"tokens": 9143779,
|
| 141 |
+
"tok_per_s": 833.2218214222122,
|
| 142 |
+
"rows_per_s": 1.534717966470355,
|
| 143 |
+
"val": {
|
| 144 |
+
"n": 1200,
|
| 145 |
+
"dropped_long": 0,
|
| 146 |
+
"noul_ce": 0.5609807372093201,
|
| 147 |
+
"noul_T": 0.8479683064850129,
|
| 148 |
+
"choice_ce": 0.6825830936431885,
|
| 149 |
+
"choice_T": 1.0268918846698019,
|
| 150 |
+
"score_ce": 1.060854196548462,
|
| 151 |
+
"score_T": 0.980406546626907,
|
| 152 |
+
"noul_auc": 0.7402538666666667,
|
| 153 |
+
"noul_acc": 0.703,
|
| 154 |
+
"noul_f1": 0.7778608825729244,
|
| 155 |
+
"noul_nll": 0.5749299364164472,
|
| 156 |
+
"noul_ece": 0.04187562750279904,
|
| 157 |
+
"noul_brier": 0.19591864012066168,
|
| 158 |
+
"noul_sharp": 0.18176335096359253,
|
| 159 |
+
"entropy": 0.5972349047660828
|
| 160 |
+
},
|
| 161 |
+
"dropped_long": 0,
|
| 162 |
+
"peak_gib": 18.237416744232178
|
| 163 |
+
}
|
| 164 |
+
}
|
meta_only/podA/e16/S1-G-d64-n16k-syn/train_log.jsonl
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"time": "20:07:29", "event": "start", "data": "data/e18/G-d64-n16k", "data_manifest": {"arm": "G-d64-n16k", "name": null, "train_sha256": "f66315602eb898d5fee95f7929164fe84183d5f3744ac956e10355b96ffee370", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 137.9488809108734}
|
| 2 |
+
{"time": "20:11:01", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 882.9253413987136, "loss": 0.5912990979850292, "entropy": 0.6006028540432453, "sharp": 0.1921953046694398}
|
| 3 |
+
{"time": "20:14:36", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 884.5994103775548, "loss": 0.6384782791137695, "entropy": 0.6375209167599678, "sharp": 0.12879459280520678}
|
| 4 |
+
{"time": "20:17:52", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 884.236199448525, "loss": 0.6149077117443085, "entropy": 0.6086066141724586, "sharp": 0.18352430406957865}
|
| 5 |
+
{"time": "20:21:09", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 885.0807233617936, "loss": 0.5936463251709938, "entropy": 0.6021706126630306, "sharp": 0.17522566951811314}
|
| 6 |
+
{"time": "20:24:40", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 884.0615737526442, "loss": 0.6008193865418434, "entropy": 0.5839069560170174, "sharp": 0.19262733403593302}
|
| 7 |
+
{"time": "20:28:07", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 883.4234056904636, "loss": 0.5711430516093969, "entropy": 0.5909330286085606, "sharp": 0.18538670800626278}
|
| 8 |
+
{"time": "20:31:25", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 882.5265377007748, "loss": 0.5924926400184631, "entropy": 0.6183340102434158, "sharp": 0.16212917305529118}
|
| 9 |
+
{"time": "20:34:53", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 882.036345732564, "loss": 0.6420952547341585, "entropy": 0.5870930403470993, "sharp": 0.20502908993512392}
|
| 10 |
+
{"time": "20:38:18", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 881.5254323274171, "loss": 0.6088168062269688, "entropy": 0.5883236806839705, "sharp": 0.18641637917608023}
|
| 11 |
+
{"time": "20:41:52", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 881.541610414559, "loss": 0.57213344424963, "entropy": 0.6258790493011475, "sharp": 0.15651467628777027}
|
| 12 |
+
{"time": "20:43:15", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5808612108230591, "noul_T": 0.733554571387925, "noul_auc": 0.7479873757404937, "noul_acc": 0.71, "noul_f1": 0.7702060221870047, "noul_nll": 0.5814999339878559, "noul_ece": 0.05849163269996644, "noul_brier": 0.19803683166812386, "noul_sharp": 0.15799014270305634, "entropy": 0.6203948855400085}
|
| 13 |
+
{"time": "20:46:40", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 850.2253360575015, "loss": 0.624946191906929, "entropy": 0.6180781293660402, "sharp": 0.15540251601487398}
|
| 14 |
+
{"time": "20:49:59", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 851.4300881188516, "loss": 0.5916736647486687, "entropy": 0.5987736731767654, "sharp": 0.18533457815647125}
|
| 15 |
+
{"time": "20:53:26", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 853.9979046549806, "loss": 0.5910882353782654, "entropy": 0.5614897347986698, "sharp": 0.2213541455566883}
|
| 16 |
+
{"time": "20:56:55", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 855.7859807630718, "loss": 0.6130129285156727, "entropy": 0.6407902166247368, "sharp": 0.138709157705307}
|
| 17 |
+
{"time": "21:00:29", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 856.8187900895524, "loss": 0.5831497944891453, "entropy": 0.5702024102210999, "sharp": 0.21758890245109797}
|
| 18 |
+
{"time": "21:04:01", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 858.6723875755686, "loss": 0.6797584854066372, "entropy": 0.6015909798443317, "sharp": 0.1792383138090372}
|
| 19 |
+
{"time": "21:07:20", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 859.7382846283366, "loss": 0.6126434225589037, "entropy": 0.5704362839460373, "sharp": 0.20829504914581776}
|
| 20 |
+
{"time": "21:10:46", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 860.9856965611436, "loss": 0.5815135128796101, "entropy": 0.6013730559498072, "sharp": 0.1772867701947689}
|
| 21 |
+
{"time": "21:14:09", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 861.9556640351949, "loss": 0.5677154529839754, "entropy": 0.6010500136762857, "sharp": 0.17544617503881454}
|
| 22 |
+
{"time": "21:17:37", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 863.0559861086, "loss": 0.6196775995194912, "entropy": 0.5943372547626495, "sharp": 0.18117377068847418}
|
| 23 |
+
{"time": "21:18:59", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5765928626060486, "noul_T": 0.8668833643945105, "noul_auc": 0.7440211979544649, "noul_acc": 0.694, "noul_f1": 0.7598116169544741, "noul_nll": 0.5793715441972017, "noul_ece": 0.060763501822948464, "noul_brier": 0.19769599788681494, "noul_sharp": 0.17913177609443665, "entropy": 0.5981926321983337}
|
| 24 |
+
{"time": "21:22:21", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 848.3971870727376, "loss": 0.6496615819633007, "entropy": 0.6177221089601517, "sharp": 0.16768693458288908}
|
| 25 |
+
{"time": "21:25:51", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 849.965286143554, "loss": 0.5621522106230259, "entropy": 0.5217941850423813, "sharp": 0.2398819848895073}
|
| 26 |
+
{"time": "21:29:17", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 851.5464462415819, "loss": 0.5804940760135651, "entropy": 0.5845111384987831, "sharp": 0.18743676226586103}
|
| 27 |
+
{"time": "21:32:51", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 852.8051509231635, "loss": 0.6010888256132603, "entropy": 0.598365779966116, "sharp": 0.1722495574504137}
|
| 28 |
+
{"time": "21:36:22", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 853.9413594579422, "loss": 0.6019518598914146, "entropy": 0.5913725383579731, "sharp": 0.19278570637106895}
|
| 29 |
+
{"time": "21:39:48", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 855.0586684587355, "loss": 0.5904597975313663, "entropy": 0.6073630601167679, "sharp": 0.1555203190073371}
|
| 30 |
+
{"time": "21:43:18", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 855.5089637363142, "loss": 0.6019733194261789, "entropy": 0.6225588396191597, "sharp": 0.15256024431437254}
|
| 31 |
+
{"time": "21:46:48", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 856.3030011013573, "loss": 0.6217922493815422, "entropy": 0.5716968141496181, "sharp": 0.20347006898373365}
|
| 32 |
+
{"time": "21:50:12", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 857.083484633367, "loss": 0.6226581037044525, "entropy": 0.6059229262173176, "sharp": 0.16864025220274925}
|
| 33 |
+
{"time": "21:53:49", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 858.0048112031558, "loss": 0.6700329408049583, "entropy": 0.6255616061389446, "sharp": 0.14370357431471348}
|
| 34 |
+
{"time": "21:55:11", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5697527527809143, "noul_T": 0.9206472336840742, "noul_auc": 0.7469241025467925, "noul_acc": 0.706, "noul_f1": 0.7789473684210526, "noul_nll": 0.5731395559236407, "noul_ece": 0.06206579971313478, "noul_brier": 0.19542217150765567, "noul_sharp": 0.19114816188812256, "entropy": 0.5833835601806641}
|
| 35 |
+
{"time": "21:58:41", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 848.0722191461795, "loss": 0.5530093535780907, "entropy": 0.5727129504084587, "sharp": 0.21449636574834585}
|
| 36 |
+
{"time": "22:02:07", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 848.9580459704615, "loss": 0.6278288885951042, "entropy": 0.5923183355480433, "sharp": 0.1856405045837164}
|
| 37 |
+
{"time": "22:05:37", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 849.997132792208, "loss": 0.5959668941795826, "entropy": 0.6114780157804489, "sharp": 0.16652793996036053}
|
| 38 |
+
{"time": "22:09:03", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 850.8813066704373, "loss": 0.5913403779268265, "entropy": 0.5878928732126951, "sharp": 0.19586333818733692}
|
| 39 |
+
{"time": "22:12:20", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 851.5896148056719, "loss": 0.6144749633967876, "entropy": 0.6340098567306995, "sharp": 0.1425577960908413}
|
| 40 |
+
{"time": "22:15:49", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 852.3721540766271, "loss": 0.6337689235806465, "entropy": 0.6213931068778038, "sharp": 0.15561622753739357}
|
| 41 |
+
{"time": "22:19:21", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 852.9091669358625, "loss": 0.6216833647340536, "entropy": 0.605075541883707, "sharp": 0.16841642558574677}
|
| 42 |
+
{"time": "22:22:47", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 853.5351449413878, "loss": 0.6208343170583248, "entropy": 0.6127814240753651, "sharp": 0.16441790759563446}
|
| 43 |
+
{"time": "22:26:13", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 854.1512758965747, "loss": 0.6115076765418053, "entropy": 0.6011828556656837, "sharp": 0.1819942109286785}
|
| 44 |
+
{"time": "22:29:57", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 854.8999840680124, "loss": 0.5834497651085258, "entropy": 0.595435980707407, "sharp": 0.17153573222458363}
|
| 45 |
+
{"time": "22:31:19", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5723874568939209, "noul_T": 0.8503527308519154, "noul_auc": 0.7481392719110226, "noul_acc": 0.7, "noul_f1": 0.7692307692307693, "noul_nll": 0.5741145680621267, "noul_ece": 0.04457561087608338, "noul_brier": 0.19547658200509466, "noul_sharp": 0.17936526238918304, "entropy": 0.596940815448761}
|
| 46 |
+
{"time": "22:34:48", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 847.5197631061586, "loss": 0.6067818440496922, "entropy": 0.6136223450303078, "sharp": 0.1588757848367095}
|
| 47 |
+
{"time": "22:38:12", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 848.3286976777166, "loss": 0.6332533210515976, "entropy": 0.6141388639807701, "sharp": 0.16970630269497633}
|
| 48 |
+
{"time": "22:41:39", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 849.1488953977947, "loss": 0.6230385284870863, "entropy": 0.6029964294284582, "sharp": 0.18261779192835093}
|
| 49 |
+
{"time": "22:45:18", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 849.667485152024, "loss": 0.6167920231819153, "entropy": 0.5847590230405331, "sharp": 0.18967749923467636}
|
| 50 |
+
{"time": "22:48:48", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 850.4017256681102, "loss": 0.5530945099890232, "entropy": 0.577253520488739, "sharp": 0.1989353485405445}
|
| 51 |
+
{"time": "22:52:13", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 850.9436427931423, "loss": 0.5726641304790974, "entropy": 0.5815538968890905, "sharp": 0.2017395105212927}
|
| 52 |
+
{"time": "22:55:28", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 851.3615917960254, "loss": 0.6311208214610815, "entropy": 0.5984008014202118, "sharp": 0.17587067931890488}
|
| 53 |
+
{"time": "22:58:58", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 851.981363693661, "loss": 0.6198936179280281, "entropy": 0.586101345717907, "sharp": 0.1912114266306162}
|
| 54 |
+
{"time": "23:02:24", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 852.6735647357523, "loss": 0.6048550494015217, "entropy": 0.6187081448733807, "sharp": 0.15867351833730936}
|
| 55 |
+
{"time": "23:06:04", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 853.3803898496399, "loss": 0.6112683303654194, "entropy": 0.6032097265124321, "sharp": 0.1715129381045699}
|
| 56 |
+
{"time": "23:07:26", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5747319459915161, "noul_T": 0.8435074373552124, "noul_auc": 0.7464346593306442, "noul_acc": 0.706, "noul_f1": 0.7706708268330733, "noul_nll": 0.5764897980690002, "noul_ece": 0.055149750828743, "noul_brier": 0.19635571144594252, "noul_sharp": 0.17810583114624023, "entropy": 0.600303590297699}
|
| 57 |
+
{"time": "23:10:24", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9143779, "tok_per_s": 833.2218214222122, "rows_per_s": 1.534717966470355, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5609807372093201, "noul_T": 0.8479683064850129, "choice_ce": 0.6825830936431885, "choice_T": 1.0268918846698019, "score_ce": 1.060854196548462, "score_T": 0.980406546626907, "noul_auc": 0.7402538666666667, "noul_acc": 0.703, "noul_f1": 0.7778608825729244, "noul_nll": 0.5749299364164472, "noul_ece": 0.04187562750279904, "noul_brier": 0.19591864012066168, "noul_sharp": 0.18176335096359253, "entropy": 0.5972349047660828}, "dropped_long": 0, "peak_gib": 18.237416744232178}
|
meta_only/podA/e16/S1-O1-near/ckpt-4000/adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 32,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.05,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 16,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"in_proj_z",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"up_proj",
|
| 42 |
+
"o_proj",
|
| 43 |
+
"v_proj",
|
| 44 |
+
"down_proj",
|
| 45 |
+
"q_proj",
|
| 46 |
+
"out_proj",
|
| 47 |
+
"k_proj"
|
| 48 |
+
],
|
| 49 |
+
"target_parameters": null,
|
| 50 |
+
"task_type": null,
|
| 51 |
+
"trainable_token_indices": null,
|
| 52 |
+
"use_bdlora": null,
|
| 53 |
+
"use_dora": false,
|
| 54 |
+
"use_qalora": false,
|
| 55 |
+
"use_rslora": false,
|
| 56 |
+
"velora_config": null
|
| 57 |
+
}
|
meta_only/podA/e16/S1-O1-near/ckpt-4000/adapter_vllm/adapter_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": {
|
| 6 |
+
"base_model_class": "Qwen3_5ForCausalLM",
|
| 7 |
+
"parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
|
| 8 |
+
},
|
| 9 |
+
"base_model_name_or_path": null,
|
| 10 |
+
"bias": "none",
|
| 11 |
+
"corda_config": null,
|
| 12 |
+
"ensure_weight_tying": false,
|
| 13 |
+
"eva_config": null,
|
| 14 |
+
"exclude_modules": null,
|
| 15 |
+
"fan_in_fan_out": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"init_lora_weights": true,
|
| 18 |
+
"kasa_config": null,
|
| 19 |
+
"layer_replication": null,
|
| 20 |
+
"layers_pattern": null,
|
| 21 |
+
"layers_to_transform": null,
|
| 22 |
+
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64.0,
|
| 24 |
+
"lora_bias": false,
|
| 25 |
+
"lora_dropout": 0.0,
|
| 26 |
+
"lora_ga_config": null,
|
| 27 |
+
"megatron_config": null,
|
| 28 |
+
"megatron_core": "megatron.core",
|
| 29 |
+
"modules_to_save": null,
|
| 30 |
+
"monteclora_config": null,
|
| 31 |
+
"peft_type": "LORA",
|
| 32 |
+
"peft_version": "0.21.0",
|
| 33 |
+
"qalora_group_size": 16,
|
| 34 |
+
"r": 32,
|
| 35 |
+
"rank_pattern": {},
|
| 36 |
+
"revision": null,
|
| 37 |
+
"target_modules": [
|
| 38 |
+
"down_proj",
|
| 39 |
+
"gate_proj",
|
| 40 |
+
"in_proj_qkv",
|
| 41 |
+
"in_proj_z",
|
| 42 |
+
"k_proj",
|
| 43 |
+
"lm_head",
|
| 44 |
+
"o_proj",
|
| 45 |
+
"out_proj",
|
| 46 |
+
"q_proj",
|
| 47 |
+
"up_proj",
|
| 48 |
+
"v_proj"
|
| 49 |
+
],
|
| 50 |
+
"target_parameters": null,
|
| 51 |
+
"task_type": null,
|
| 52 |
+
"trainable_token_indices": null,
|
| 53 |
+
"use_bdlora": null,
|
| 54 |
+
"use_dora": false,
|
| 55 |
+
"use_qalora": false,
|
| 56 |
+
"use_rslora": false,
|
| 57 |
+
"velora_config": null
|
| 58 |
+
}
|