Hugging Face's logo

LEAP-LAB-KUS
/
jevkt-checkpoints

PEFT
Safetensors
knowledge-tracing
system-one
typed-decisions
lora
calibration
Model card Files
xet
 Community
Unggi commited on
Commit
e1eb80d
·
verified ·
1 Parent(s): 50b689a

backup e16/S2-grpo-nostd (Pod B)

Browse files
Files changed (19) hide show
  1. e16/S2-grpo-nostd/ckpt-16000/adapter/README.md +206 -0
  2. e16/S2-grpo-nostd/ckpt-16000/adapter/adapter_config.json +57 -0
  3. e16/S2-grpo-nostd/ckpt-16000/adapter/adapter_model.safetensors +3 -0
  4. e16/S2-grpo-nostd/ckpt-16000/adapter_vllm/adapter_config.json +58 -0
  5. e16/S2-grpo-nostd/ckpt-16000/adapter_vllm/adapter_model.safetensors +3 -0
  6. e16/S2-grpo-nostd/ckpt-16000/adapter_vllm/decision_head.json +99 -0
  7. e16/S2-grpo-nostd/ckpt-16000/calibration.json +17 -0
  8. e16/S2-grpo-nostd/ckpt-16000/head.safetensors +3 -0
  9. e16/S2-grpo-nostd/ckpt-16000/judge_config.json +160 -0
  10. e16/S2-grpo-nostd/ckpt-4000/adapter/README.md +206 -0
  11. e16/S2-grpo-nostd/ckpt-4000/adapter/adapter_config.json +57 -0
  12. e16/S2-grpo-nostd/ckpt-4000/adapter/adapter_model.safetensors +3 -0
  13. e16/S2-grpo-nostd/ckpt-4000/adapter_vllm/adapter_config.json +58 -0
  14. e16/S2-grpo-nostd/ckpt-4000/adapter_vllm/adapter_model.safetensors +3 -0
  15. e16/S2-grpo-nostd/ckpt-4000/adapter_vllm/decision_head.json +99 -0
  16. e16/S2-grpo-nostd/ckpt-4000/calibration.json +17 -0
  17. e16/S2-grpo-nostd/ckpt-4000/head.safetensors +3 -0
  18. e16/S2-grpo-nostd/ckpt-4000/judge_config.json +160 -0
  19. e16/S2-grpo-nostd/train_log.jsonl +58 -0
e16/S2-grpo-nostd/ckpt-16000/adapter/README.md ADDED
@@ -0,0 +1,206 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: /workspace/jev_kt/models/hf/JEV-9B
3
+ library_name: peft
4
+ tags:
5
+ - base_model:adapter:/workspace/jev_kt/models/hf/JEV-9B
6
+ - lora
7
+ - transformers
8
+ ---
9
+
10
+ # Model Card for Model ID
11
+
12
+ <!-- Provide a quick summary of what the model is/does. -->
13
+
14
+
15
+
16
+ ## Model Details
17
+
18
+ ### Model Description
19
+
20
+ <!-- Provide a longer summary of what this model is. -->
21
+
22
+
23
+
24
+ - **Developed by:** [More Information Needed]
25
+ - **Funded by [optional]:** [More Information Needed]
26
+ - **Shared by [optional]:** [More Information Needed]
27
+ - **Model type:** [More Information Needed]
28
+ - **Language(s) (NLP):** [More Information Needed]
29
+ - **License:** [More Information Needed]
30
+ - **Finetuned from model [optional]:** [More Information Needed]
31
+
32
+ ### Model Sources [optional]
33
+
34
+ <!-- Provide the basic links for the model. -->
35
+
36
+ - **Repository:** [More Information Needed]
37
+ - **Paper [optional]:** [More Information Needed]
38
+ - **Demo [optional]:** [More Information Needed]
39
+
40
+ ## Uses
41
+
42
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
43
+
44
+ ### Direct Use
45
+
46
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
47
+
48
+ [More Information Needed]
49
+
50
+ ### Downstream Use [optional]
51
+
52
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
53
+
54
+ [More Information Needed]
55
+
56
+ ### Out-of-Scope Use
57
+
58
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
59
+
60
+ [More Information Needed]
61
+
62
+ ## Bias, Risks, and Limitations
63
+
64
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
65
+
66
+ [More Information Needed]
67
+
68
+ ### Recommendations
69
+
70
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
71
+
72
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
73
+
74
+ ## How to Get Started with the Model
75
+
76
+ Use the code below to get started with the model.
77
+
78
+ [More Information Needed]
79
+
80
+ ## Training Details
81
+
82
+ ### Training Data
83
+
84
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
85
+
86
+ [More Information Needed]
87
+
88
+ ### Training Procedure
89
+
90
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
91
+
92
+ #### Preprocessing [optional]
93
+
94
+ [More Information Needed]
95
+
96
+
97
+ #### Training Hyperparameters
98
+
99
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
100
+
101
+ #### Speeds, Sizes, Times [optional]
102
+
103
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
104
+
105
+ [More Information Needed]
106
+
107
+ ## Evaluation
108
+
109
+ <!-- This section describes the evaluation protocols and provides the results. -->
110
+
111
+ ### Testing Data, Factors & Metrics
112
+
113
+ #### Testing Data
114
+
115
+ <!-- This should link to a Dataset Card if possible. -->
116
+
117
+ [More Information Needed]
118
+
119
+ #### Factors
120
+
121
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
122
+
123
+ [More Information Needed]
124
+
125
+ #### Metrics
126
+
127
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
128
+
129
+ [More Information Needed]
130
+
131
+ ### Results
132
+
133
+ [More Information Needed]
134
+
135
+ #### Summary
136
+
137
+
138
+
139
+ ## Model Examination [optional]
140
+
141
+ <!-- Relevant interpretability work for the model goes here -->
142
+
143
+ [More Information Needed]
144
+
145
+ ## Environmental Impact
146
+
147
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
148
+
149
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
150
+
151
+ - **Hardware Type:** [More Information Needed]
152
+ - **Hours used:** [More Information Needed]
153
+ - **Cloud Provider:** [More Information Needed]
154
+ - **Compute Region:** [More Information Needed]
155
+ - **Carbon Emitted:** [More Information Needed]
156
+
157
+ ## Technical Specifications [optional]
158
+
159
+ ### Model Architecture and Objective
160
+
161
+ [More Information Needed]
162
+
163
+ ### Compute Infrastructure
164
+
165
+ [More Information Needed]
166
+
167
+ #### Hardware
168
+
169
+ [More Information Needed]
170
+
171
+ #### Software
172
+
173
+ [More Information Needed]
174
+
175
+ ## Citation [optional]
176
+
177
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
178
+
179
+ **BibTeX:**
180
+
181
+ [More Information Needed]
182
+
183
+ **APA:**
184
+
185
+ [More Information Needed]
186
+
187
+ ## Glossary [optional]
188
+
189
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
190
+
191
+ [More Information Needed]
192
+
193
+ ## More Information [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Authors [optional]
198
+
199
+ [More Information Needed]
200
+
201
+ ## Model Card Contact
202
+
203
+ [More Information Needed]
204
+ ### Framework versions
205
+
206
+ - PEFT 0.21.2
e16/S2-grpo-nostd/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_qkv",
39
+ "out_proj",
40
+ "v_proj",
41
+ "in_proj_z",
42
+ "q_proj",
43
+ "o_proj",
44
+ "up_proj",
45
+ "k_proj",
46
+ "down_proj",
47
+ "gate_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
+ }
e16/S2-grpo-nostd/ckpt-16000/adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95bd12e37c70e39ea06fe6a7362fac6ec20a4727a2418fd19f452f2b83f972b9
3
+ size 160486456
e16/S2-grpo-nostd/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
+ }
e16/S2-grpo-nostd/ckpt-16000/adapter_vllm/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef2c460e3cb47859981ca2b17ba019d69ad7cf58005d2e25fb7fa4905147d6cb
3
+ size 176641832
e16/S2-grpo-nostd/ckpt-16000/adapter_vllm/decision_head.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bias": [
3
+ 0.0021994407288730145,
4
+ 0.0014450125163421035,
5
+ -0.0025878134183585644,
6
+ -0.0012615364976227283,
7
+ -0.00025007399381138384,
8
+ 0.0019558551721274853,
9
+ 0.0013559526996687055,
10
+ -0.0023348911199718714,
11
+ -0.0004082925443071872,
12
+ -0.0002078723773593083,
13
+ -1.2957353646925185e-05,
14
+ 0.0001676026004133746,
15
+ 0.0005569425993598998,
16
+ -0.001412136247381568,
17
+ -0.0028663526754826307,
18
+ -0.005061559844762087,
19
+ -0.00593328895047307,
20
+ -0.004319756291806698,
21
+ -0.004922541789710522,
22
+ -0.0049067651852965355,
23
+ -0.003943523392081261,
24
+ -0.003319827374070883,
25
+ -0.005581587553024292,
26
+ -0.0034480984322726727
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
+ }
e16/S2-grpo-nostd/ckpt-16000/calibration.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 1,
3
+ "per_kind": {
4
+ "noul": 1.3104006628864093,
5
+ "choice": 1.0,
6
+ "score": 1.0
7
+ },
8
+ "fit": {
9
+ "noul": {
10
+ "n": 2000,
11
+ "T": 1.3104006628864093,
12
+ "ce_before": 0.539025068283081,
13
+ "ce_after": 0.5299394726753235
14
+ }
15
+ },
16
+ "source": "models/e16/S2-grpo-nostd/ckpt-16000"
17
+ }
e16/S2-grpo-nostd/ckpt-16000/head.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:32f69c205829911168d5c06305b17aaa0ff06246129e8bbcca4618f66011404b
3
+ size 393464
e16/S2-grpo-nostd/ckpt-16000/judge_config.json ADDED
@@ -0,0 +1,160 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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/S2-grpo-nostd/ckpt-16000",
103
+ "softcap": null,
104
+ "e16": {
105
+ "data": "data/e15/F1-pool",
106
+ "data_manifest": {
107
+ "arm": "F1-pool",
108
+ "name": "F1-pool",
109
+ "train_sha256": "dad16f94a5659768b254a6c12538c01f9c64cd43d574a3435bcc70c9c1bac130",
110
+ "n_train": 64000
111
+ },
112
+ "init": "released",
113
+ "objective": "grpo-nostd",
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": null,
124
+ "replay_frac": 0.0,
125
+ "replay_sha256": null,
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": 0,
139
+ "step": 500,
140
+ "tokens": 10544036,
141
+ "tok_per_s": 797.4121095335993,
142
+ "rows_per_s": 1.2100294183087554,
143
+ "val": {
144
+ "n": 2000,
145
+ "dropped_long": 0,
146
+ "noul_ce": 0.539025068283081,
147
+ "noul_T": 1.3104006628864093,
148
+ "noul_auc": 0.7753633867792973,
149
+ "noul_acc": 0.7335,
150
+ "noul_f1": 0.8045471213788046,
151
+ "noul_nll": 0.5390251027175691,
152
+ "noul_ece": 0.05838637080788614,
153
+ "noul_brier": 0.1792133274542628,
154
+ "noul_sharp": 0.2777736783027649,
155
+ "entropy": 0.4669798016548157
156
+ },
157
+ "dropped_long": 0,
158
+ "peak_gib": 18.236032962799072
159
+ }
160
+ }
e16/S2-grpo-nostd/ckpt-4000/adapter/README.md ADDED
@@ -0,0 +1,206 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: /workspace/jev_kt/models/hf/JEV-9B
3
+ library_name: peft
4
+ tags:
5
+ - base_model:adapter:/workspace/jev_kt/models/hf/JEV-9B
6
+ - lora
7
+ - transformers
8
+ ---
9
+
10
+ # Model Card for Model ID
11
+
12
+ <!-- Provide a quick summary of what the model is/does. -->
13
+
14
+
15
+
16
+ ## Model Details
17
+
18
+ ### Model Description
19
+
20
+ <!-- Provide a longer summary of what this model is. -->
21
+
22
+
23
+
24
+ - **Developed by:** [More Information Needed]
25
+ - **Funded by [optional]:** [More Information Needed]
26
+ - **Shared by [optional]:** [More Information Needed]
27
+ - **Model type:** [More Information Needed]
28
+ - **Language(s) (NLP):** [More Information Needed]
29
+ - **License:** [More Information Needed]
30
+ - **Finetuned from model [optional]:** [More Information Needed]
31
+
32
+ ### Model Sources [optional]
33
+
34
+ <!-- Provide the basic links for the model. -->
35
+
36
+ - **Repository:** [More Information Needed]
37
+ - **Paper [optional]:** [More Information Needed]
38
+ - **Demo [optional]:** [More Information Needed]
39
+
40
+ ## Uses
41
+
42
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
43
+
44
+ ### Direct Use
45
+
46
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
47
+
48
+ [More Information Needed]
49
+
50
+ ### Downstream Use [optional]
51
+
52
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
53
+
54
+ [More Information Needed]
55
+
56
+ ### Out-of-Scope Use
57
+
58
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
59
+
60
+ [More Information Needed]
61
+
62
+ ## Bias, Risks, and Limitations
63
+
64
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
65
+
66
+ [More Information Needed]
67
+
68
+ ### Recommendations
69
+
70
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
71
+
72
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
73
+
74
+ ## How to Get Started with the Model
75
+
76
+ Use the code below to get started with the model.
77
+
78
+ [More Information Needed]
79
+
80
+ ## Training Details
81
+
82
+ ### Training Data
83
+
84
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
85
+
86
+ [More Information Needed]
87
+
88
+ ### Training Procedure
89
+
90
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
91
+
92
+ #### Preprocessing [optional]
93
+
94
+ [More Information Needed]
95
+
96
+
97
+ #### Training Hyperparameters
98
+
99
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
100
+
101
+ #### Speeds, Sizes, Times [optional]
102
+
103
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
104
+
105
+ [More Information Needed]
106
+
107
+ ## Evaluation
108
+
109
+ <!-- This section describes the evaluation protocols and provides the results. -->
110
+
111
+ ### Testing Data, Factors & Metrics
112
+
113
+ #### Testing Data
114
+
115
+ <!-- This should link to a Dataset Card if possible. -->
116
+
117
+ [More Information Needed]
118
+
119
+ #### Factors
120
+
121
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
122
+
123
+ [More Information Needed]
124
+
125
+ #### Metrics
126
+
127
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
128
+
129
+ [More Information Needed]
130
+
131
+ ### Results
132
+
133
+ [More Information Needed]
134
+
135
+ #### Summary
136
+
137
+
138
+
139
+ ## Model Examination [optional]
140
+
141
+ <!-- Relevant interpretability work for the model goes here -->
142
+
143
+ [More Information Needed]
144
+
145
+ ## Environmental Impact
146
+
147
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
148
+
149
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
150
+
151
+ - **Hardware Type:** [More Information Needed]
152
+ - **Hours used:** [More Information Needed]
153
+ - **Cloud Provider:** [More Information Needed]
154
+ - **Compute Region:** [More Information Needed]
155
+ - **Carbon Emitted:** [More Information Needed]
156
+
157
+ ## Technical Specifications [optional]
158
+
159
+ ### Model Architecture and Objective
160
+
161
+ [More Information Needed]
162
+
163
+ ### Compute Infrastructure
164
+
165
+ [More Information Needed]
166
+
167
+ #### Hardware
168
+
169
+ [More Information Needed]
170
+
171
+ #### Software
172
+
173
+ [More Information Needed]
174
+
175
+ ## Citation [optional]
176
+
177
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
178
+
179
+ **BibTeX:**
180
+
181
+ [More Information Needed]
182
+
183
+ **APA:**
184
+
185
+ [More Information Needed]
186
+
187
+ ## Glossary [optional]
188
+
189
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
190
+
191
+ [More Information Needed]
192
+
193
+ ## More Information [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Authors [optional]
198
+
199
+ [More Information Needed]
200
+
201
+ ## Model Card Contact
202
+
203
+ [More Information Needed]
204
+ ### Framework versions
205
+
206
+ - PEFT 0.21.2
e16/S2-grpo-nostd/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_qkv",
39
+ "out_proj",
40
+ "v_proj",
41
+ "in_proj_z",
42
+ "q_proj",
43
+ "o_proj",
44
+ "up_proj",
45
+ "k_proj",
46
+ "down_proj",
47
+ "gate_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
+ }
e16/S2-grpo-nostd/ckpt-4000/adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7188b2b2c1a75ff771d53c2228ee7a8f5128159fe3aea3a1fb76db6f009af782
3
+ size 160486456
e16/S2-grpo-nostd/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
+ }
e16/S2-grpo-nostd/ckpt-4000/adapter_vllm/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fe7fd6add71f444583adc2328b7cb6398f2e44f69482cebf967f666570ae4d9
3
+ size 176641832
e16/S2-grpo-nostd/ckpt-4000/adapter_vllm/decision_head.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bias": [
3
+ 0.001987096853554249,
4
+ 0.0011478476226329803,
5
+ -0.0025878134183585644,
6
+ -0.0012615364976227283,
7
+ -0.00025007399381138384,
8
+ 0.0019558551721274853,
9
+ 0.0013559526996687055,
10
+ -0.0023348911199718714,
11
+ -0.0004082925443071872,
12
+ -0.0002078723773593083,
13
+ -1.2957353646925185e-05,
14
+ 0.0001676026004133746,
15
+ 0.0005569425993598998,
16
+ -0.001412136247381568,
17
+ -0.0028663526754826307,
18
+ -0.005061559844762087,
19
+ -0.00593328895047307,
20
+ -0.004319756291806698,
21
+ -0.004922541789710522,
22
+ -0.0049067651852965355,
23
+ -0.003943523392081261,
24
+ -0.003319827374070883,
25
+ -0.005581587553024292,
26
+ -0.0034480984322726727
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
+ }
e16/S2-grpo-nostd/ckpt-4000/calibration.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 1,
3
+ "per_kind": {
4
+ "noul": 1.0112425650460184,
5
+ "choice": 1.0,
6
+ "score": 1.0
7
+ },
8
+ "fit": {
9
+ "noul": {
10
+ "n": 2000,
11
+ "T": 1.0112425650460184,
12
+ "ce_before": 0.5447059869766235,
13
+ "ce_after": 0.5446915030479431
14
+ }
15
+ },
16
+ "source": "models/e16/S2-grpo-nostd/ckpt-4000"
17
+ }
e16/S2-grpo-nostd/ckpt-4000/head.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a61cbd6c498c61d862cf3b70c9ac1c16c57867dc3c8b2fec21a42b42bfa31ec
3
+ size 393464
e16/S2-grpo-nostd/ckpt-4000/judge_config.json ADDED
@@ -0,0 +1,160 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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/S2-grpo-nostd/ckpt-4000",
103
+ "softcap": null,
104
+ "e16": {
105
+ "data": "data/e15/F1-pool",
106
+ "data_manifest": {
107
+ "arm": "F1-pool",
108
+ "name": "F1-pool",
109
+ "train_sha256": "dad16f94a5659768b254a6c12538c01f9c64cd43d574a3435bcc70c9c1bac130",
110
+ "n_train": 64000
111
+ },
112
+ "init": "released",
113
+ "objective": "grpo-nostd",
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": null,
124
+ "replay_frac": 0.0,
125
+ "replay_sha256": null,
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-4000",
137
+ "rows_seen": 4000,
138
+ "replay_seen": 0,
139
+ "step": 125,
140
+ "tokens": 2631181,
141
+ "tok_per_s": 754.5500632067781,
142
+ "rows_per_s": 1.1470895568398345,
143
+ "val": {
144
+ "n": 2000,
145
+ "dropped_long": 0,
146
+ "noul_ce": 0.5447059869766235,
147
+ "noul_T": 1.0112425650460184,
148
+ "noul_auc": 0.7598141505083899,
149
+ "noul_acc": 0.7255,
150
+ "noul_f1": 0.8017334777898159,
151
+ "noul_nll": 0.5447060198821128,
152
+ "noul_ece": 0.027148934215307244,
153
+ "noul_brier": 0.1831792210588921,
154
+ "noul_sharp": 0.2391430139541626,
155
+ "entropy": 0.5421568751335144
156
+ },
157
+ "dropped_long": 0,
158
+ "peak_gib": 18.23348903656006
159
+ }
160
+ }
e16/S2-grpo-nostd/train_log.jsonl ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time": "17:59:41", "event": "start", "data": "data/e15/F1-pool", "data_manifest": {"arm": "F1-pool", "name": "F1-pool", "train_sha256": "dad16f94a5659768b254a6c12538c01f9c64cd43d574a3435bcc70c9c1bac130", "n_train": 64000}, "init": "released", "objective": "grpo-nostd", "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": null, "replay_frac": 0.0, "replay_sha256": null, "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": 76.19867300987244}
2
+ {"time": "18:03:40", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 884.4544615973111, "loss": 1.1641532182693481e-09, "entropy": 0.3600083850324154, "sharp": 0.34307756274938583, "r_mean": 0.5268827577431997, "r_std": 0.14494017884135246, "adv_abs": 0.1128927730023861, "frac_lowstd": 0.027777778605620067}
3
+ {"time": "18:07:43", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 885.0653804271913, "loss": 3.026798367500305e-09, "entropy": 0.5488521195948124, "sharp": 0.23003558441996574, "r_mean": 0.5858212198529925, "r_std": 0.21750972419977188, "adv_abs": 0.1653505734034947, "frac_lowstd": 0.0}
4
+ {"time": "18:11:41", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 883.1231160038733, "loss": 1.6298145055770874e-09, "entropy": 0.3594680242240429, "sharp": 0.3525789678096771, "r_mean": 0.580526909657887, "r_std": 0.11532950560961451, "adv_abs": 0.0862714515200683, "frac_lowstd": 0.05238095351627895}
5
+ {"time": "18:15:43", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 885.2186599395266, "loss": 4.190951585769653e-09, "entropy": 0.2485016086138785, "sharp": 0.4102237615734339, "r_mean": 0.23109945548432215, "r_std": 0.07861641448523317, "adv_abs": 0.05814485957047769, "frac_lowstd": 0.15714285842009953}
6
+ {"time": "18:19:42", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 886.0182290957656, "loss": 1.979060471057892e-09, "entropy": 0.37756368424743414, "sharp": 0.3504131566733122, "r_mean": 0.4071050371442522, "r_std": 0.14521778441433394, "adv_abs": 0.11443485798580307, "frac_lowstd": 0.0}
7
+ {"time": "18:23:40", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 886.2769747346898, "loss": 1.6298145055770874e-09, "entropy": 0.4882803726941347, "sharp": 0.27374129369854927, "r_mean": 0.5262709536722728, "r_std": 0.15644189502511704, "adv_abs": 0.12036370486021042, "frac_lowstd": 0.0}
8
+ {"time": "18:27:36", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 885.2032660673434, "loss": -3.7834979593753815e-09, "entropy": 0.5390893332660198, "sharp": 0.22214827965945005, "r_mean": 0.45258228161505293, "r_std": 0.2516068773610251, "adv_abs": 0.19478647197995866, "frac_lowstd": 0.0}
9
+ {"time": "18:31:35", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 885.7663375081723, "loss": 8.149072527885437e-10, "entropy": 0.31276444531977177, "sharp": 0.3642871379852295, "r_mean": 0.7114843939031873, "r_std": 0.1047235799155065, "adv_abs": 0.080613616188722, "frac_lowstd": 0.14897959572928293}
10
+ {"time": "18:35:35", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 885.0090647151733, "loss": 8.149072527885437e-10, "entropy": 0.3364601107314229, "sharp": 0.34918440878391266, "r_mean": 0.575337644134249, "r_std": 0.10669690184295177, "adv_abs": 0.0821004259799208, "frac_lowstd": 0.11785714541162763}
11
+ {"time": "18:39:32", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 885.4898870519502, "loss": -4.0745362639427185e-09, "entropy": 0.3527836985886097, "sharp": 0.3443656526505947, "r_mean": 0.5247749225901706, "r_std": 0.13196771166154317, "adv_abs": 0.10115048289299011, "frac_lowstd": 0.0}
12
+ {"time": "18:41:18", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.760796308517456, "noul_T": 2.7467957675184738, "noul_auc": 0.7704545050569687, "noul_acc": 0.672, "noul_f1": 0.7248322147651006, "noul_nll": 0.7607963000042364, "noul_ece": 0.1867298910841346, "noul_brier": 0.24107829594827324, "noul_sharp": 0.35169145464897156, "entropy": 0.35047993063926697}
13
+ {"time": "18:45:16", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 852.5265691069005, "loss": -5.820766091346741e-10, "entropy": 0.15239583628135733, "sharp": 0.4353483021259308, "r_mean": 0.48013102502695154, "r_std": 0.05648724087763769, "adv_abs": 0.046465574630669186, "frac_lowstd": 0.5253968302692685}
14
+ {"time": "18:49:06", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 854.9122905597551, "loss": 3.259629011154175e-09, "entropy": 0.3314599636942148, "sharp": 0.36463984474539757, "r_mean": 0.4178845818553652, "r_std": 0.11807165188448769, "adv_abs": 0.08893829158374242, "frac_lowstd": 0.0}
15
+ {"time": "18:57:51", "event": "ckpt", "tag": "ckpt-4000", "rows_seen": 4000, "replay_seen": 0, "step": 125, "tokens": 2631181, "tok_per_s": 754.5500632067781, "rows_per_s": 1.1470895568398345, "val": {"n": 2000, "dropped_long": 0, "noul_ce": 0.5447059869766235, "noul_T": 1.0112425650460184, "noul_auc": 0.7598141505083899, "noul_acc": 0.7255, "noul_f1": 0.8017334777898159, "noul_nll": 0.5447060198821128, "noul_ece": 0.027148934215307244, "noul_brier": 0.1831792210588921, "noul_sharp": 0.2391430139541626, "entropy": 0.5421568751335144}, "dropped_long": 0, "peak_gib": 18.23348903656006}
16
+ {"time": "18:59:51", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 758.826940870863, "loss": 3.4924596548080444e-09, "entropy": 0.5845713298767805, "sharp": 0.2007088791579008, "r_mean": 0.5733937110219683, "r_std": 0.21517409809998103, "adv_abs": 0.17115230964762823, "frac_lowstd": 0.0}
17
+ {"time": "19:03:48", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 766.7469209903467, "loss": 1.1641532182693481e-09, "entropy": 0.4589529503136873, "sharp": 0.2872612848877907, "r_mean": 0.44099410836185726, "r_std": 0.1733246615954808, "adv_abs": 0.13490553945302963, "frac_lowstd": 0.0}
18
+ {"time": "19:07:47", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 773.8201644797298, "loss": -5.413312464952469e-09, "entropy": 0.5307577345520258, "sharp": 0.2309715561568737, "r_mean": 0.5797681340149471, "r_std": 0.1803292515022414, "adv_abs": 0.13750122593981878, "frac_lowstd": 0.0}
19
+ {"time": "19:11:47", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 780.3699898557727, "loss": 6.868503987789154e-09, "entropy": 0.5565783903002739, "sharp": 0.21629841066896915, "r_mean": 0.5385217467943827, "r_std": 0.25154653936624527, "adv_abs": 0.189613809188207, "frac_lowstd": 0.0}
20
+ {"time": "19:15:46", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 786.0096662170888, "loss": -1.1408701539039612e-08, "entropy": 0.37103007826954126, "sharp": 0.3308442961424589, "r_mean": 0.6875255874225071, "r_std": 0.10993586347571441, "adv_abs": 0.08233889830963952, "frac_lowstd": 0.04421768869672503}
21
+ {"time": "19:19:36", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 790.7769069939661, "loss": -3.0850060284137726e-09, "entropy": 0.46601253747940063, "sharp": 0.269111474044621, "r_mean": 0.45283375432093936, "r_std": 0.18254188696543375, "adv_abs": 0.14007492860158285, "frac_lowstd": 0.0}
22
+ {"time": "19:23:30", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 795.6844564001739, "loss": -5.587935447692871e-09, "entropy": 0.5188713259994984, "sharp": 0.2551375897601247, "r_mean": 0.45337540762765066, "r_std": 0.2093526359115328, "adv_abs": 0.1598963322384017, "frac_lowstd": 0.028571428997176036}
23
+ {"time": "19:27:37", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 799.8735189844542, "loss": -4.0745362639427185e-09, "entropy": 0.5082225564401597, "sharp": 0.22050108388066292, "r_mean": 0.6753702632018498, "r_std": 0.166827213684363, "adv_abs": 0.1306292244366237, "frac_lowstd": 0.08843537739345006}
24
+ {"time": "19:29:23", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.6028103828430176, "noul_T": 1.6965604390111058, "noul_auc": 0.7239019534652233, "noul_acc": 0.686, "noul_f1": 0.796895213454075, "noul_nll": 0.6028103750068694, "noul_ece": 0.08544198536872864, "noul_brier": 0.20089477414545331, "noul_sharp": 0.24649374186992645, "entropy": 0.47840994596481323}
25
+ {"time": "19:33:21", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 788.6498938686934, "loss": -5.995389074087143e-09, "entropy": 0.47226636204868555, "sharp": 0.2648441642522812, "r_mean": 0.5826202588421958, "r_std": 0.156695800700358, "adv_abs": 0.12175089014428002, "frac_lowstd": 0.017857142857142856}
26
+ {"time": "19:37:11", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 792.7571238520203, "loss": -2.6775524020195007e-09, "entropy": 0.2960859062150121, "sharp": 0.38094215653836727, "r_mean": 0.6634682863950729, "r_std": 0.10834534683575232, "adv_abs": 0.08590894440809886, "frac_lowstd": 0.2759259318312009}
27
+ {"time": "19:41:10", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 796.7733774266163, "loss": 8.032657206058502e-09, "entropy": 0.48289890959858894, "sharp": 0.26723550632596016, "r_mean": 0.5556483396462032, "r_std": 0.20538687705993652, "adv_abs": 0.15557628550699779, "frac_lowstd": 0.07142857142857142}
28
+ {"time": "19:44:58", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 800.264751392517, "loss": 2.7939677238464355e-09, "entropy": 0.388722681440413, "sharp": 0.31518605910241604, "r_mean": 0.6629531724112374, "r_std": 0.14829746234629834, "adv_abs": 0.1177349526967321, "frac_lowstd": 0.14557823538780212}
29
+ {"time": "19:48:55", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 803.4568884414577, "loss": 1.1641532182693481e-09, "entropy": 0.3980801859870553, "sharp": 0.298974359408021, "r_mean": 0.8300266095570156, "r_std": 0.10686029367414969, "adv_abs": 0.08366367087832519, "frac_lowstd": 0.09523809594767434}
30
+ {"time": "19:52:44", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 806.3337509532315, "loss": -5.005858838558197e-09, "entropy": 0.3227362767793238, "sharp": 0.3588506653904915, "r_mean": 0.636544440473829, "r_std": 0.10457484791654029, "adv_abs": 0.07890124140041214, "frac_lowstd": 0.1357142882687705}
31
+ {"time": "19:56:40", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 808.9988036464595, "loss": -5.3551048040390015e-09, "entropy": 0.43946995586156845, "sharp": 0.290117260068655, "r_mean": 0.5732823063929876, "r_std": 0.16074063753088316, "adv_abs": 0.12638770788908005, "frac_lowstd": 0.15158730496962866}
32
+ {"time": "20:00:38", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 811.7679348590942, "loss": -2.0954757928848267e-09, "entropy": 0.3961673453450203, "sharp": 0.3205057755112648, "r_mean": 0.5788945513112205, "r_std": 0.1635388976761273, "adv_abs": 0.12097743685756411, "frac_lowstd": 0.020408164177622114}
33
+ {"time": "20:04:38", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 814.4227327440462, "loss": -7.799826562404633e-09, "entropy": 0.3683847263455391, "sharp": 0.3342200331389904, "r_mean": 0.7833065560885838, "r_std": 0.112129514105618, "adv_abs": 0.08431503655655044, "frac_lowstd": 0.07142857355730874}
34
+ {"time": "20:08:33", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 816.6104511310798, "loss": 7.8580342233181e-09, "entropy": 0.37657117121852934, "sharp": 0.3202393501996994, "r_mean": 0.5817821781550135, "r_std": 0.13586779959898973, "adv_abs": 0.1057657642023904, "frac_lowstd": 0.07142857355730874}
35
+ {"time": "20:10:19", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.6698617339134216, "noul_T": 2.3871365741174793, "noul_auc": 0.7654241987948596, "noul_acc": 0.738, "noul_f1": 0.8241610738255034, "noul_nll": 0.6698618221593788, "noul_ece": 0.13359233009815216, "noul_brier": 0.19880243973260622, "noul_sharp": 0.35738325119018555, "entropy": 0.32139748334884644}
36
+ {"time": "20:14:24", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 807.9062258496411, "loss": 1.280568540096283e-09, "entropy": 0.403143085539341, "sharp": 0.3126984592527151, "r_mean": 0.4497241973876953, "r_std": 0.14574840664863586, "adv_abs": 0.11035739364368576, "frac_lowstd": 0.023809524519102915}
37
+ {"time": "20:18:20", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 810.2350281755163, "loss": -3.026798367500305e-09, "entropy": 0.3747909180819988, "sharp": 0.3466624617576599, "r_mean": 0.7044756838253566, "r_std": 0.12996902848993028, "adv_abs": 0.09777420439890452, "frac_lowstd": 0.05238095351627895}
38
+ {"time": "20:22:16", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 812.2307561275342, "loss": 8.149072527885437e-10, "entropy": 0.428047239780426, "sharp": 0.3008122220635414, "r_mean": 0.44331876157472533, "r_std": 0.19420302659273148, "adv_abs": 0.15216707810759544, "frac_lowstd": 0.0}
39
+ {"time": "20:26:19", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 814.3464723857722, "loss": 2.0954757928848267e-09, "entropy": 0.48615321703255177, "sharp": 0.26783706061542034, "r_mean": 0.5853140290294375, "r_std": 0.18209448776074819, "adv_abs": 0.14268102709736144, "frac_lowstd": 0.0}
40
+ {"time": "20:30:17", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 816.4080842711388, "loss": -2.270098775625229e-09, "entropy": 0.5014967103488743, "sharp": 0.25410042330622673, "r_mean": 0.6703737452626228, "r_std": 0.183393782004714, "adv_abs": 0.1360396547242999, "frac_lowstd": 0.0}
41
+ {"time": "20:34:10", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 818.1406283211787, "loss": -1.280568540096283e-09, "entropy": 0.42276160791516304, "sharp": 0.2960072997957468, "r_mean": 0.7215448447636196, "r_std": 0.13142795621284417, "adv_abs": 0.09912447737795967, "frac_lowstd": 0.023809524519102915}
42
+ {"time": "20:38:05", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 819.9439919429092, "loss": -4.190951585769653e-09, "entropy": 0.47549560479819775, "sharp": 0.2780580874532461, "r_mean": 0.6109173127583095, "r_std": 0.15389065550906317, "adv_abs": 0.12034723854490689, "frac_lowstd": 0.0}
43
+ {"time": "20:42:02", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 821.6902201565252, "loss": 2.0954757928848267e-09, "entropy": 0.4314619340002537, "sharp": 0.3040547538548708, "r_mean": 0.601762843983514, "r_std": 0.14216666535607406, "adv_abs": 0.11224054917693138, "frac_lowstd": 0.16666666950498307}
44
+ {"time": "20:45:59", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 823.28473720808, "loss": 3.725290298461914e-09, "entropy": 0.34322142926976085, "sharp": 0.34586818516254425, "r_mean": 0.7992831553731646, "r_std": 0.07801502341005419, "adv_abs": 0.05918294031705175, "frac_lowstd": 0.09404762089252472}
45
+ {"time": "20:49:58", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 824.7001176360717, "loss": -7.566995918750763e-09, "entropy": 0.3913815291598439, "sharp": 0.3262783456593752, "r_mean": 0.8334476947784424, "r_std": 0.10199685326369945, "adv_abs": 0.07620461052283645, "frac_lowstd": 0.015625}
46
+ {"time": "20:51:43", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5847235321998596, "noul_T": 1.7622834987463807, "noul_auc": 0.7674683161805222, "noul_acc": 0.74, "noul_f1": 0.8238482384823849, "noul_nll": 0.5847235770970582, "noul_ece": 0.09478526896238326, "noul_brier": 0.18792842278693475, "noul_sharp": 0.31628167629241943, "entropy": 0.40062204003334045}
47
+ {"time": "20:55:45", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 817.8837184926062, "loss": -1.280568540096283e-09, "entropy": 0.3781249690800905, "sharp": 0.3313314877450466, "r_mean": 0.7598850280046463, "r_std": 0.12635319431622824, "adv_abs": 0.09286383042732875, "frac_lowstd": 0.10714286069075267}
48
+ {"time": "20:59:43", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 819.533296146445, "loss": 1.1641532182693481e-09, "entropy": 0.4792353864759207, "sharp": 0.26845137495547533, "r_mean": 0.5625139235385826, "r_std": 0.1832280574100358, "adv_abs": 0.13906531035900116, "frac_lowstd": 0.0}
49
+ {"time": "21:03:45", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 820.9122814144591, "loss": 4.6566128730773926e-09, "entropy": 0.33765628561377525, "sharp": 0.34808861650526524, "r_mean": 0.515512724007879, "r_std": 0.18532496158565795, "adv_abs": 0.14590600372425147, "frac_lowstd": 0.12857143155166081}
50
+ {"time": "21:07:37", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 822.3719401536135, "loss": -5.238689482212067e-09, "entropy": 0.3542978633195162, "sharp": 0.35381273925304413, "r_mean": 0.7106397698322932, "r_std": 0.10294737666845322, "adv_abs": 0.07960239425301552, "frac_lowstd": 0.08492063730955124}
51
+ {"time": "21:11:37", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 823.672725064141, "loss": -6.752088665962219e-09, "entropy": 0.4770170683041215, "sharp": 0.27162476629018784, "r_mean": 0.681754823241915, "r_std": 0.14957339423043386, "adv_abs": 0.11463062465190887, "frac_lowstd": 0.07142857142857142}
52
+ {"time": "21:15:34", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 825.0048141438365, "loss": 4.6566128730773926e-09, "entropy": 0.4035720834508538, "sharp": 0.32604275457561016, "r_mean": 0.5468903386167118, "r_std": 0.1379139288328588, "adv_abs": 0.10519787669181824, "frac_lowstd": 0.07142857142857142}
53
+ {"time": "21:19:35", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 826.4143804961512, "loss": 1.1059455573558807e-09, "entropy": 0.47471519000828266, "sharp": 0.25355912931263447, "r_mean": 0.6629931075232369, "r_std": 0.15895063230501755, "adv_abs": 0.12447464466094971, "frac_lowstd": 0.020408164177622114}
54
+ {"time": "21:23:30", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 827.473522381079, "loss": 5.238689482212067e-09, "entropy": 0.4484249968081713, "sharp": 0.29879135824739933, "r_mean": 0.6261742015679678, "r_std": 0.15800587398310503, "adv_abs": 0.12214213858048122, "frac_lowstd": 0.020833333333333332}
55
+ {"time": "21:27:28", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 828.7489212387869, "loss": -5.005858838558197e-09, "entropy": 0.4283724557608366, "sharp": 0.2976568881422281, "r_mean": 0.6017972528934479, "r_std": 0.12359725258180074, "adv_abs": 0.09773840169821467, "frac_lowstd": 0.0}
56
+ {"time": "21:31:26", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 829.9203717209649, "loss": -1.0477378964424133e-09, "entropy": 0.4702608473598957, "sharp": 0.27212344110012054, "r_mean": 0.7544927100340525, "r_std": 0.1572697882850965, "adv_abs": 0.12058918923139572, "frac_lowstd": 0.027777778605620067}
57
+ {"time": "21:33:11", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5487236380577087, "noul_T": 1.3525088368861407, "noul_auc": 0.7676816153859828, "noul_acc": 0.738, "noul_f1": 0.8059259259259259, "noul_nll": 0.5487236229386181, "noul_ece": 0.08295815402269366, "noul_brier": 0.18214486865307591, "noul_sharp": 0.27488580346107483, "entropy": 0.46956557035446167}
58
+ {"time": "21:40:06", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 0, "step": 500, "tokens": 10544036, "tok_per_s": 797.4121095335993, "rows_per_s": 1.2100294183087554, "val": {"n": 2000, "dropped_long": 0, "noul_ce": 0.539025068283081, "noul_T": 1.3104006628864093, "noul_auc": 0.7753633867792973, "noul_acc": 0.7335, "noul_f1": 0.8045471213788046, "noul_nll": 0.5390251027175691, "noul_ece": 0.05838637080788614, "noul_brier": 0.1792133274542628, "noul_sharp": 0.2777736783027649, "entropy": 0.4669798016548157}, "dropped_long": 0, "peak_gib": 18.236032962799072}