mehmet1899 commited on
Commit
87afdd0
·
verified ·
1 Parent(s): e86ed85

Upload final Llama 3.2 3B Instruct NL2SQL LoRA adapter

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: meta-llama/Llama-3.2-3B-Instruct
3
+ library_name: peft
4
+ pipeline_tag: text-generation
5
+ language:
6
+ - en
7
+ tags:
8
+ - peft
9
+ - lora
10
+ - text-to-sql
11
+ - nl2sql
12
+ - spider
13
+ - sqlite
14
+ ---
15
+
16
+ # Llama 3.2 3B Instruct NL2SQL LoRA
17
+
18
+ This repository contains the final LoRA adapter for the Llama 3.2 3B Instruct model line used in a master's thesis project on local large language models for NL2SQL generation.
19
+
20
+ The adapter generates SQLite queries from natural-language questions and a relational database schema. It must be loaded together with the corresponding base model.
21
+
22
+ ## Base model
23
+
24
+ - Model: `meta-llama/Llama-3.2-3B-Instruct`
25
+ - Revision: `0cb88a4f764b7a12671c53f0838cd831a0843b95`
26
+ - Access to the base model may require acceptance of Meta's license terms on Hugging Face.
27
+
28
+ ## Adapter
29
+
30
+ - Method: LoRA supervised fine-tuning
31
+ - LoRA rank: 8
32
+ - LoRA alpha: 16
33
+ - LoRA dropout: 0.05
34
+ - Target modules: all suitable linear modules
35
+ - Quantization during training: none
36
+ - Maximum training sequence length: 2,048 tokens
37
+ - Best checkpoint: `checkpoint-509`
38
+ - The published root adapter corresponds to the selected best checkpoint.
39
+
40
+ SHA-256 of `adapter_model.safetensors`:
41
+
42
+ `fcd4241f7a2e8e0388f13f0dd9517486cbee43fc3169c983a54e7b716c0e502d`
43
+
44
+ ## Training configuration
45
+
46
+ - Training examples: 25,000
47
+ - Spider Train examples: 6,960
48
+ - SQL Create Context examples: 18,040
49
+ - Validation set: MixedVal2500-v2
50
+ - Validation examples: 2,500
51
+ - Learning rate: `1e-4`
52
+ - Scheduler: constant
53
+ - Train batch size: 2
54
+ - Gradient accumulation steps: 4
55
+ - Effective batch size: 8
56
+ - Seed: 42
57
+ - Maximum epochs: 5
58
+ - Early stopping patience: 2
59
+ - Early stopping threshold: 0.001
60
+ - Precision: FP16
61
+ - Gradient checkpointing: enabled
62
+ - Attention implementation: FlashAttention 2
63
+
64
+ Spider Dev was not used for training, validation, early stopping, or checkpoint selection.
65
+
66
+ ## Evaluation
67
+
68
+ The final adapter was evaluated on all 1,032 Spider Dev cases.
69
+
70
+ Zero-shot evaluation:
71
+
72
+ - Execution Match Accuracy: 61.05% (630/1,032)
73
+ - Execution Success Rate: 86.82% (896/1,032)
74
+ - Maximum input length: 2,048 tokens
75
+ - Maximum generated tokens: 256
76
+
77
+ The corresponding starting model achieved an Execution Match Accuracy of 55.04% (568/1,032) under the same zero-shot evaluation condition.
78
+
79
+ ## Intended prompt behavior
80
+
81
+ The model is instructed to return only a valid SQLite query:
82
+
83
+ - no explanation
84
+ - no Markdown
85
+ - no comments
86
+ - no unnecessary tables or columns
87
+ - only `SELECT` or `WITH` queries
88
+ - output terminated with a semicolon
89
+
90
+ The native Llama chat template and the exact project-specific prompt construction are documented in the accompanying GitHub repository.
91
+
92
+ ## Loading
93
+
94
+ ~~~python
95
+ import torch
96
+ from peft import PeftModel
97
+ from transformers import AutoModelForCausalLM, AutoTokenizer
98
+
99
+ base_model_id = "meta-llama/Llama-3.2-3B-Instruct"
100
+ adapter_id = "mehmet1899/llama32-3b-instruct-nl2sql-lora"
101
+
102
+ tokenizer = AutoTokenizer.from_pretrained(
103
+ base_model_id,
104
+ revision="0cb88a4f764b7a12671c53f0838cd831a0843b95",
105
+ )
106
+
107
+ model = AutoModelForCausalLM.from_pretrained(
108
+ base_model_id,
109
+ revision="0cb88a4f764b7a12671c53f0838cd831a0843b95",
110
+ torch_dtype=torch.float16,
111
+ device_map="auto",
112
+ )
113
+
114
+ model = PeftModel.from_pretrained(model, adapter_id)
115
+ model.eval()
116
+ ~~~
117
+
118
+ ## Reproducibility
119
+
120
+ Code, training and evaluation configurations, environment information, run manifests, and result summaries are available at:
121
+
122
+ `https://github.com/md181099/nl2sql-masterthesis`
123
+
124
+ The files `training_metadata.json`, `training_history.csv`, and `training_history.jsonl` provide additional training provenance.
125
+
126
+ ## Limitations
127
+
128
+ - The adapter was evaluated primarily on the Spider benchmark and SQLite databases.
129
+ - Performance on other database systems or unseen schema conventions is not guaranteed.
130
+ - Access to the base model is governed by Meta's model license.
131
+ - Execution Match depends on the database contents and the execution-based evaluation procedure.
132
+ - The model may still generate invalid, incomplete, or semantically incorrect SQL.
133
+ - The adapter should not be used to execute unrestricted queries against production databases without validation and access controls.
SHA256SUMS ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ 8c7c14a41632a1ab3564fa25521f8737237795b5dcae7730abfe0ab4aca88480 adapter_config.json
2
+ fcd4241f7a2e8e0388f13f0dd9517486cbee43fc3169c983a54e7b716c0e502d adapter_model.safetensors
3
+ bcfbdd3b4b74206c9a61aff857b91e32cfe6f54dd329eb134eccf566b2be8c41 README.md
4
+ 5816fce10444e03c2e9ee1ef8a4a1ea61ae7e69e438613f3b17b69d0426223a4 chat_template.jinja
5
+ 58d954db45152000b3596db19f22648213abd8a65893610380e527b7c87802b6 tokenizer_config.json
6
+ 6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b tokenizer.json
7
+ f5365d265fe999047a78dc0ebb94d8deb563eabee0ccdadc4de149f280ceb7ea training_metadata.json
8
+ 5947a547d51049836504823fbc88b1b621376614e798654b0bf82240d3373f69 training_history.csv
9
+ 2e40d8195699ea3b1a158430527f4bbd8468c922c004670d94f31cfc4e77c71b training_history.jsonl
adapter_config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alora_invocation_tokens": null,
3
+ "alpha_pattern": {},
4
+ "arrow_config": null,
5
+ "auto_mapping": null,
6
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B-Instruct",
7
+ "bias": "none",
8
+ "corda_config": null,
9
+ "ensure_weight_tying": false,
10
+ "eva_config": null,
11
+ "exclude_modules": null,
12
+ "fan_in_fan_out": false,
13
+ "inference_mode": true,
14
+ "init_lora_weights": true,
15
+ "layer_replication": null,
16
+ "layers_pattern": null,
17
+ "layers_to_transform": null,
18
+ "loftq_config": {},
19
+ "lora_alpha": 16,
20
+ "lora_bias": false,
21
+ "lora_dropout": 0.05,
22
+ "megatron_config": null,
23
+ "megatron_core": "megatron.core",
24
+ "modules_to_save": null,
25
+ "peft_type": "LORA",
26
+ "peft_version": "0.18.1",
27
+ "qalora_group_size": 16,
28
+ "r": 8,
29
+ "rank_pattern": {},
30
+ "revision": null,
31
+ "target_modules": [
32
+ "up_proj",
33
+ "q_proj",
34
+ "o_proj",
35
+ "gate_proj",
36
+ "k_proj",
37
+ "v_proj",
38
+ "down_proj"
39
+ ],
40
+ "target_parameters": null,
41
+ "task_type": "CAUSAL_LM",
42
+ "trainable_token_indices": null,
43
+ "use_dora": false,
44
+ "use_qalora": false,
45
+ "use_rslora": false
46
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcd4241f7a2e8e0388f13f0dd9517486cbee43fc3169c983a54e7b716c0e502d
3
+ size 48679352
chat_template.jinja ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {{- bos_token }}
2
+ {%- if custom_tools is defined %}
3
+ {%- set tools = custom_tools %}
4
+ {%- endif %}
5
+ {%- if not tools_in_user_message is defined %}
6
+ {%- set tools_in_user_message = true %}
7
+ {%- endif %}
8
+ {%- if not date_string is defined %}
9
+ {%- if strftime_now is defined %}
10
+ {%- set date_string = strftime_now("%d %b %Y") %}
11
+ {%- else %}
12
+ {%- set date_string = "26 Jul 2024" %}
13
+ {%- endif %}
14
+ {%- endif %}
15
+ {%- if not tools is defined %}
16
+ {%- set tools = none %}
17
+ {%- endif %}
18
+
19
+ {#- This block extracts the system message, so we can slot it into the right place. #}
20
+ {%- if messages[0]['role'] == 'system' %}
21
+ {%- set system_message = messages[0]['content']|trim %}
22
+ {%- set messages = messages[1:] %}
23
+ {%- else %}
24
+ {%- set system_message = "" %}
25
+ {%- endif %}
26
+
27
+ {#- System message #}
28
+ {{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
29
+ {%- if tools is not none %}
30
+ {{- "Environment: ipython\n" }}
31
+ {%- endif %}
32
+ {{- "Cutting Knowledge Date: December 2023\n" }}
33
+ {{- "Today Date: " + date_string + "\n\n" }}
34
+ {%- if tools is not none and not tools_in_user_message %}
35
+ {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}
36
+ {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
37
+ {{- "Do not use variables.\n\n" }}
38
+ {%- for t in tools %}
39
+ {{- t | tojson(indent=4) }}
40
+ {{- "\n\n" }}
41
+ {%- endfor %}
42
+ {%- endif %}
43
+ {{- system_message }}
44
+ {{- "<|eot_id|>" }}
45
+
46
+ {#- Custom tools are passed in a user message with some extra guidance #}
47
+ {%- if tools_in_user_message and not tools is none %}
48
+ {#- Extract the first user message so we can plug it in here #}
49
+ {%- if messages | length != 0 %}
50
+ {%- set first_user_message = messages[0]['content']|trim %}
51
+ {%- set messages = messages[1:] %}
52
+ {%- else %}
53
+ {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
54
+ {%- endif %}
55
+ {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
56
+ {{- "Given the following functions, please respond with a JSON for a function call " }}
57
+ {{- "with its proper arguments that best answers the given prompt.\n\n" }}
58
+ {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
59
+ {{- "Do not use variables.\n\n" }}
60
+ {%- for t in tools %}
61
+ {{- t | tojson(indent=4) }}
62
+ {{- "\n\n" }}
63
+ {%- endfor %}
64
+ {{- first_user_message + "<|eot_id|>"}}
65
+ {%- endif %}
66
+
67
+ {%- for message in messages %}
68
+ {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
69
+ {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }}
70
+ {%- elif 'tool_calls' in message %}
71
+ {%- if not message.tool_calls|length == 1 %}
72
+ {{- raise_exception("This model only supports single tool-calls at once!") }}
73
+ {%- endif %}
74
+ {%- set tool_call = message.tool_calls[0].function %}
75
+ {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
76
+ {{- '{"name": "' + tool_call.name + '", ' }}
77
+ {{- '"parameters": ' }}
78
+ {{- tool_call.arguments | tojson }}
79
+ {{- "}" }}
80
+ {{- "<|eot_id|>" }}
81
+ {%- elif message.role == "tool" or message.role == "ipython" %}
82
+ {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}
83
+ {%- if message.content is mapping or message.content is iterable %}
84
+ {{- message.content | tojson }}
85
+ {%- else %}
86
+ {{- message.content }}
87
+ {%- endif %}
88
+ {{- "<|eot_id|>" }}
89
+ {%- endif %}
90
+ {%- endfor %}
91
+ {%- if add_generation_prompt %}
92
+ {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
93
+ {%- endif %}
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
3
+ size 17209920
tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<|begin_of_text|>",
4
+ "clean_up_tokenization_spaces": true,
5
+ "eos_token": "<|eot_id|>",
6
+ "is_local": false,
7
+ "local_files_only": false,
8
+ "model_input_names": [
9
+ "input_ids",
10
+ "attention_mask"
11
+ ],
12
+ "model_max_length": 131072,
13
+ "pad_token": "<|eot_id|>",
14
+ "tokenizer_class": "TokenizersBackend"
15
+ }
training_history.csv ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ step,epoch,loss,eval_loss,grad_norm,learning_rate,entropy,num_tokens,mean_token_accuracy,train_runtime,timestamp,elapsed_seconds
2
+ 10,0.0196753566158,2.21839561462,,0.997185707092,0.0001,1.51200933158,158517,0.617804196477,,2026-07-14T18:44:51.763785+00:00,40.3268301521
3
+ 20,0.0393507132317,1.0074344635,,1.03053867817,0.0001,1.0494144395,315386,0.796838776767,,2026-07-14T18:45:31.035504+00:00,79.598546861
4
+ 30,0.0590260698475,0.544442272186,,0.422507971525,0.0001,0.548911824077,472899,0.898587629199,,2026-07-14T18:46:11.591751+00:00,120.154795361
5
+ 40,0.0787014264634,0.445441532135,,0.18211427331,0.0001,0.455530235171,629242,0.916029234231,,2026-07-14T18:46:50.703015+00:00,159.266057246
6
+ 50,0.0983767830792,0.451950931549,,0.175900384784,0.0001,0.460615222156,785551,0.911731639504,,2026-07-14T18:47:30.680216+00:00,199.243261393
7
+ 60,0.118052139695,0.427107477188,,0.218168973923,0.0001,0.443916998804,942530,0.914710514247,,2026-07-14T18:48:10.859048+00:00,239.422092887
8
+ 70,0.137727496311,0.397407078743,,0.257529675961,0.0001,0.416817698628,1100964,0.920619755983,,2026-07-14T18:48:51.257122+00:00,279.820165421
9
+ 80,0.157402852927,0.35759768486,,0.660029649734,0.0001,0.384742297977,1257044,0.924064183235,,2026-07-14T18:49:30.413227+00:00,318.97627022
10
+ 90,0.177078209543,0.35855255127,,0.264365077019,0.0001,0.391139079258,1414621,0.923739472032,,2026-07-14T18:50:10.999452+00:00,359.562497077
11
+ 100,0.196753566158,0.317103385925,,0.273194640875,0.0001,0.336241504923,1571856,0.932000268996,,2026-07-14T18:50:50.553821+00:00,399.116867291
12
+ 110,0.216428922774,0.315398597717,,0.224106535316,0.0001,0.332903936133,1729714,0.933503301442,,2026-07-14T18:51:31.107001+00:00,439.670047054
13
+ 120,0.23610427939,0.283270096779,,0.243651434779,0.0001,0.303971963003,1886650,0.938405682147,,2026-07-14T18:52:10.758208+00:00,479.321254092
14
+ 130,0.255779636006,0.277483892441,,0.242042064667,0.0001,0.299890926853,2042273,0.940673132241,,2026-07-14T18:52:50.714737+00:00,519.277783558
15
+ 140,0.275454992622,0.265112280846,,0.265905737877,0.0001,0.288989692181,2198932,0.942814520001,,2026-07-14T18:53:30.145592+00:00,558.708636761
16
+ 150,0.295130349238,0.254311847687,,0.307109236717,0.0001,0.265480846912,2355875,0.945721656084,,2026-07-14T18:54:10.566345+00:00,599.12936811
17
+ 160,0.314805705853,0.23861758709,,0.344656556845,0.0001,0.259159233049,2512427,0.948875762522,,2026-07-14T18:54:50.199052+00:00,638.762097128
18
+ 170,0.334481062469,0.230272626877,,0.261819094419,0.0001,0.242634578422,2668770,0.950659604371,,2026-07-14T18:55:30.244558+00:00,678.807603146
19
+ 180,0.354156419085,0.214873051643,,0.348735779524,0.0001,0.232795054093,2825377,0.954010221362,,2026-07-14T18:56:09.792053+00:00,718.355098992
20
+ 190,0.373831775701,0.206166386604,,0.238673120737,0.0001,0.220631313324,2982619,0.955823786557,,2026-07-14T18:56:50.404094+00:00,758.96713975
21
+ 200,0.393507132317,0.214299988747,,0.259331971407,0.0001,0.234532970004,3138939,0.954085548222,,2026-07-14T18:57:29.415292+00:00,797.978338393
22
+ 210,0.413182488933,0.199846196175,,0.242847457528,0.0001,0.212221397832,3294561,0.95643684268,,2026-07-14T18:58:09.741501+00:00,838.304546847
23
+ 220,0.432857845548,0.204166269302,,0.306981146336,0.0001,0.214753072709,3449991,0.956322589517,,2026-07-14T18:58:49.004264+00:00,877.567307537
24
+ 230,0.452533202164,0.201543879509,,0.249639883637,0.0001,0.216364040598,3608785,0.956451144814,,2026-07-14T18:59:29.554834+00:00,918.117879928
25
+ 240,0.47220855878,0.215132331848,,0.24389693141,0.0001,0.223825023696,3766845,0.953913928568,,2026-07-14T19:00:09.144558+00:00,957.707604611
26
+ 250,0.491883915396,0.200074696541,,0.322559267282,0.0001,0.216750285402,3923032,0.957909616828,,2026-07-14T19:00:49.451045+00:00,998.014089783
27
+ 260,0.511559272012,0.198160004616,,0.223266273737,0.0001,0.210631896928,4080458,0.957292881608,,2026-07-14T19:01:28.755248+00:00,1037.31829346
28
+ 270,0.531234628628,0.206540870667,,0.225164785981,0.0001,0.218400148116,4236536,0.955662851036,,2026-07-14T19:02:09.002715+00:00,1077.56576105
29
+ 280,0.550909985243,0.203662276268,,0.211283221841,0.0001,0.212336610816,4391657,0.955616168678,,2026-07-14T19:02:48.202244+00:00,1116.7652876
30
+ 290,0.570585341859,0.198586702347,,0.29632806778,0.0001,0.207531724125,4548276,0.957449166477,,2026-07-14T19:03:28.562368+00:00,1157.12541444
31
+ 300,0.590260698475,0.1769338727,,0.200340017676,0.0001,0.189514789172,4706349,0.961175942421,,2026-07-14T19:04:08.011571+00:00,1196.57461494
32
+ 310,0.609936055091,0.180926573277,,0.217008456588,0.0001,0.196929321997,4863585,0.96009849906,,2026-07-14T19:04:48.498705+00:00,1237.06175009
33
+ 320,0.629611411707,0.180410706997,,0.236014738679,0.0001,0.18719286155,5022342,0.960626663268,,2026-07-14T19:05:28.044393+00:00,1276.60743859
34
+ 330,0.649286768323,0.187188267708,,0.259057343006,0.0001,0.197457621247,5177117,0.958998320997,,2026-07-14T19:06:07.805697+00:00,1316.3687436
35
+ 340,0.668962124939,0.203676509857,,0.230080202222,0.0001,0.207781772502,5334589,0.956454566121,,2026-07-14T19:06:47.379224+00:00,1355.94226928
36
+ 350,0.688637481554,0.174643802643,,0.283878415823,0.0001,0.185111483,5491264,0.961374931037,,2026-07-14T19:07:27.523734+00:00,1396.08677869
37
+ 360,0.70831283817,0.179219591618,,0.24345164001,0.0001,0.196479166858,5648766,0.96104246974,,2026-07-14T19:08:07.102058+00:00,1435.66510349
38
+ 370,0.727988194786,0.172455787659,,0.217018336058,0.0001,0.181602787226,5805950,0.961948390305,,2026-07-14T19:08:47.480806+00:00,1476.04385117
39
+ 380,0.747663551402,0.17482407093,,0.328721493483,0.0001,0.180027039908,5962530,0.961873859167,,2026-07-14T19:09:26.926417+00:00,1515.48946346
40
+ 390,0.767338908018,0.162100684643,,0.200087547302,0.0001,0.176972093433,6119998,0.964109444618,,2026-07-14T19:10:07.159783+00:00,1555.72282812
41
+ 400,0.787014264634,0.183642709255,,0.196165248752,0.0001,0.189363373816,6276791,0.960221962631,,2026-07-14T19:10:46.801680+00:00,1595.36472525
42
+ 410,0.806689621249,0.171787154675,,0.245862558484,0.0001,0.182321485505,6433516,0.961653226614,,2026-07-14T19:11:27.245900+00:00,1635.80894558
43
+ 420,0.826364977865,0.182455682755,,0.325130581856,0.0001,0.191438901238,6590564,0.959084931016,,2026-07-14T19:12:06.417007+00:00,1674.98005193
44
+ 430,0.846040334481,0.169191372395,,0.207633405924,0.0001,0.178573282529,6747353,0.963288144767,,2026-07-14T19:12:46.362587+00:00,1714.92563345
45
+ 440,0.865715691097,0.180155730247,,0.206554561853,0.0001,0.18681143187,6902907,0.960263897479,,2026-07-14T19:13:25.393102+00:00,1753.95614771
46
+ 450,0.885391047713,0.171831595898,,0.30511072278,0.0001,0.181087859161,7059033,0.962434618175,,2026-07-14T19:14:05.571981+00:00,1794.13502563
47
+ 460,0.905066404329,0.180643391609,,0.196387454867,0.0001,0.188525561802,7216633,0.959945641458,,2026-07-14T19:14:44.847650+00:00,1833.4106959
48
+ 470,0.924741760944,0.193449866772,,0.19179905951,0.0001,0.196602920815,7373319,0.95812100023,,2026-07-14T19:15:26.044097+00:00,1874.60714292
49
+ 480,0.94441711756,0.177313435078,,0.155064806342,0.0001,0.188440449163,7530283,0.961460134387,,2026-07-14T19:16:06.338099+00:00,1914.90114522
50
+ 490,0.964092474176,0.180543804169,,0.233374238014,0.0001,0.187788728997,7687237,0.960137434304,,2026-07-14T19:16:46.621608+00:00,1955.18465457
51
+ 500,0.983767830792,0.180359601974,,0.180976226926,0.0001,0.188501783088,7844660,0.96057779789,,2026-07-14T19:17:25.826646+00:00,1994.38969213
52
+ 509,1,,0.480833500624,,,,,,,2026-07-14T19:19:37.994260+00:00,2126.55730501
53
+ 510,1.00196753566,0.192923069,,0.169916674495,0.0001,0.198045271675,7988594,0.957474518467,,2026-07-14T19:19:42.166818+00:00,2130.72986391
54
+ 520,1.02164289228,0.170230865479,,0.194753885269,0.0001,0.178488218039,8144390,0.961938048899,,2026-07-14T19:20:21.674456+00:00,2170.23750172
55
+ 530,1.04131824889,0.156556117535,,0.168322607875,0.0001,0.165742422082,8300636,0.965336105227,,2026-07-14T19:21:01.779702+00:00,2210.3427483
56
+ 540,1.06099360551,0.160328769684,,0.186233431101,0.0001,0.168560723402,8458859,0.96414308995,,2026-07-14T19:21:41.348379+00:00,2249.91142491
57
+ 550,1.08066896212,0.169278275967,,0.182504862547,0.0001,0.173748625629,8615963,0.963006226718,,2026-07-14T19:22:21.535345+00:00,2290.09839124
58
+ 560,1.10034431874,0.15963177681,,0.230876639485,0.0001,0.170929611102,8772742,0.964167003334,,2026-07-14T19:23:01.254806+00:00,2329.81785137
59
+ 570,1.12001967536,0.157126617432,,0.179838463664,0.0001,0.16031345129,8930284,0.964271454513,,2026-07-14T19:23:41.828440+00:00,2370.3914869
60
+ 580,1.13969503197,0.166155254841,,0.223575741053,0.0001,0.175235996023,9088333,0.963411171734,,2026-07-14T19:24:21.736483+00:00,2410.29952893
61
+ 590,1.15937038859,0.157854199409,,0.207558766007,0.0001,0.166331799515,9244311,0.964486956596,,2026-07-14T19:25:02.048624+00:00,2450.61167042
62
+ 600,1.1790457452,0.162935483456,,0.173033863306,0.0001,0.167879271507,9402415,0.963171319664,,2026-07-14T19:25:42.017999+00:00,2490.58104435
63
+ 610,1.19872110182,0.17940980196,,0.200073346496,0.0001,0.188126351219,9556833,0.960648794472,,2026-07-14T19:26:22.097129+00:00,2530.6601727
64
+ 620,1.21839645844,0.165559697151,,0.176193639636,0.0001,0.171528098173,9714854,0.962894457579,,2026-07-14T19:27:01.449098+00:00,2570.01214384
65
+ 630,1.23807181505,0.192057442665,,0.249208107591,0.0001,0.198998957127,9871371,0.957532766461,,2026-07-14T19:27:41.929207+00:00,2610.49225355
66
+ 640,1.25774717167,0.16299200058,,0.19560906291,0.0001,0.172278838418,10028132,0.964198562503,,2026-07-14T19:28:21.258453+00:00,2649.82149892
67
+ 650,1.27742252828,0.182843160629,,0.250791400671,0.0001,0.185129570402,10183998,0.959990593791,,2026-07-14T19:29:01.072607+00:00,2689.6356517
68
+ 660,1.2970978849,0.156658506393,,0.173409461975,0.0001,0.165276159532,10338966,0.96491754353,,2026-07-14T19:29:39.981675+00:00,2728.5447211
69
+ 670,1.31677324152,0.156918954849,,0.2154648453,0.0001,0.165447768662,10494154,0.964841979742,,2026-07-14T19:30:20.537538+00:00,2769.10058474
70
+ 680,1.33644859813,0.168604540825,,0.202249929309,0.0001,0.174681568705,10650728,0.962572240829,,2026-07-14T19:30:59.771538+00:00,2808.33458337
71
+ 690,1.35612395475,0.179817044735,,0.180585950613,0.0001,0.184981219843,10808084,0.960407233238,,2026-07-14T19:31:40.333487+00:00,2848.89653316
72
+ 700,1.37579931136,0.17117511034,,0.189327448606,0.0001,0.179342577234,10965357,0.962315896153,,2026-07-14T19:32:20.034892+00:00,2888.597938
73
+ 710,1.39547466798,0.163115501404,,0.14550216496,0.0001,0.170794568304,11122970,0.963731876016,,2026-07-14T19:33:00.055825+00:00,2928.61887098
74
+ 720,1.41515002459,0.166223096848,,0.197181358933,0.0001,0.174024736509,11279748,0.962751935422,,2026-07-14T19:33:39.118755+00:00,2967.68180075
75
+ 730,1.43482538121,0.171106719971,,0.232188314199,0.0001,0.175083556771,11437649,0.962013469636,,2026-07-14T19:34:19.410048+00:00,3007.9730946
76
+ 740,1.45450073783,0.159269690514,,0.179410859942,0.0001,0.167735586967,11595537,0.964913637936,,2026-07-14T19:34:59.021640+00:00,3047.58468635
77
+ 750,1.47417609444,0.168200385571,,0.155719816685,0.0001,0.175214201398,11753054,0.962682126462,,2026-07-14T19:35:39.483000+00:00,3088.04604582
78
+ 760,1.49385145106,0.14346704483,,0.221360266209,0.0001,0.154037842248,11911753,0.968222369254,,2026-07-14T19:36:18.999097+00:00,3127.56214216
79
+ 770,1.51352680767,0.161464035511,,0.167245700955,0.0001,0.165353834908,12069142,0.963533814251,,2026-07-14T19:36:59.202932+00:00,3167.76597772
80
+ 780,1.53320216429,0.156456899643,,0.17038269341,0.0001,0.165936999302,12224271,0.96456181556,,2026-07-14T19:37:37.739912+00:00,3206.30295782
81
+ 790,1.55287752091,0.162909126282,,0.194951727986,0.0001,0.166183103435,12380914,0.963615015149,,2026-07-14T19:38:17.951802+00:00,3246.51484807
82
+ 800,1.57255287752,0.162587237358,,0.175505697727,0.0001,0.170294051804,12538265,0.963853444159,,2026-07-14T19:38:57.359451+00:00,3285.92249702
83
+ 810,1.59222823414,0.16598303318,,0.182322755456,0.0001,0.171620354988,12694472,0.962860387564,,2026-07-14T19:39:37.707953+00:00,3326.2709987
84
+ 820,1.61190359075,0.166602897644,,0.162591278553,0.0001,0.173711701855,12852241,0.962746998668,,2026-07-14T19:40:17.543353+00:00,3366.10638842
85
+ 830,1.63157894737,0.173879754543,,0.179233163595,0.0001,0.180813023262,13010313,0.960821950436,,2026-07-14T19:40:58.733570+00:00,3407.29661541
86
+ 840,1.65125430398,0.175928378105,,0.155069187284,0.0001,0.180705674551,13167919,0.960775198042,,2026-07-14T19:41:38.127878+00:00,3446.69092196
87
+ 850,1.6709296606,0.156295228004,,0.187289491296,0.0001,0.163406656217,13326698,0.964838838577,,2026-07-14T19:42:18.752868+00:00,3487.3159135
88
+ 860,1.69060501722,0.166846621037,,0.162943869829,0.0001,0.173019717634,13483225,0.962671551108,,2026-07-14T19:42:57.972132+00:00,3526.53517718
89
+ 870,1.71028037383,0.175749349594,,0.174129873514,0.0001,0.181726143695,13639265,0.960963216424,,2026-07-14T19:43:37.800757+00:00,3566.36380294
90
+ 880,1.72995573045,0.147835624218,,0.167158782482,0.0001,0.154400302656,13797228,0.966271243989,,2026-07-14T19:44:17.466713+00:00,3606.02975898
91
+ 890,1.74963108706,0.16417195797,,0.196655958891,0.0001,0.170779196545,13954300,0.963016964495,,2026-07-14T19:44:57.876500+00:00,3646.43954604
92
+ 900,1.76930644368,0.156272256374,,0.16130194068,0.0001,0.164454833046,14111213,0.964544859529,,2026-07-14T19:45:37.195299+00:00,3685.75834587
93
+ 910,1.7889818003,0.163049316406,,0.136302247643,0.0001,0.169371592253,14267408,0.963653604686,,2026-07-14T19:46:17.611788+00:00,3726.17483362
94
+ 920,1.80865715691,0.147612285614,,0.16987170279,0.0001,0.15369244311,14424482,0.966288872063,,2026-07-14T19:46:57.069799+00:00,3765.63284427
95
+ 930,1.82833251353,0.151848042011,,0.155094489455,0.0001,0.158277668338,14581898,0.965862759948,,2026-07-14T19:47:37.633555+00:00,3806.19659956
96
+ 940,1.84800787014,0.161747670174,,0.176779091358,0.0001,0.167998809647,14739682,0.963628257811,,2026-07-14T19:48:16.925203+00:00,3845.48824898
97
+ 950,1.86768322676,0.175212359428,,0.186388149858,0.0001,0.179522324633,14897444,0.961017121375,,2026-07-14T19:48:57.156443+00:00,3885.71948934
98
+ 960,1.88735858337,0.158414626122,,0.143919214606,0.0001,0.163319427148,15053687,0.964056003094,,2026-07-14T19:49:36.454119+00:00,3925.01716572
99
+ 970,1.90703393999,0.176569092274,,0.212138965726,0.0001,0.18215863388,15210316,0.960795244575,,2026-07-14T19:50:16.690489+00:00,3965.25353464
100
+ 980,1.92670929661,0.158397960663,,0.177053377032,0.0001,0.166588350572,15367289,0.964338231087,,2026-07-14T19:50:55.920110+00:00,4004.48315612
101
+ 990,1.94638465322,0.172202885151,,0.198846206069,0.0001,0.173680017237,15519567,0.962071706355,,2026-07-14T19:51:34.604574+00:00,4043.16762026
102
+ 1000,1.96606000984,0.135841083527,,0.173086538911,0.0001,0.145909713674,15677796,0.968235512078,,2026-07-14T19:52:13.740978+00:00,4082.30402441
103
+ 1010,1.98573536645,0.169320774078,,0.182816982269,0.0001,0.173335280083,15834854,0.962027944624,,2026-07-14T19:52:54.082399+00:00,4122.64544415
104
+ 1018,2,,0.503623247147,,,,,,,2026-07-14T19:55:03.085112+00:00,4251.64815671
105
+ 1020,2.00393507132,0.156750667095,,0.21632951498,0.0001,0.174239883048,15980450,0.96212148022,,2026-07-14T19:55:11.229820+00:00,4259.79286485
106
+ 1030,2.02361042794,0.16210129261,,0.189968839288,0.0001,0.16485402789,16135755,0.963425774872,,2026-07-14T19:55:51.178359+00:00,4299.74140436
107
+ 1040,2.04328578455,0.151982796192,,0.186117559671,0.0001,0.164776344784,16293622,0.96450523138,,2026-07-14T19:56:30.672001+00:00,4339.23504656
108
+ 1050,2.06296114117,0.163860476017,,0.191996067762,0.0001,0.16807874972,16449011,0.962708187103,,2026-07-14T19:57:10.452525+00:00,4379.01557092
109
+ 1060,2.08263649779,0.172380006313,,0.209191724658,0.0001,0.179409135506,16606544,0.961136734486,,2026-07-14T19:57:49.831844+00:00,4418.39488819
110
+ 1070,2.1023118544,0.141785228252,,0.14348590374,0.0001,0.150052259676,16762655,0.967867693305,,2026-07-14T19:58:29.903173+00:00,4458.46621657
111
+ 1080,2.12198721102,0.158943665028,,0.173628717661,0.0001,0.165021989681,16918189,0.963940601051,,2026-07-14T19:59:09.353150+00:00,4497.9161958
112
+ 1090,2.14166256763,0.161983144283,,0.170995950699,0.0001,0.169115608837,17074380,0.9631606251,,2026-07-14T19:59:49.097296+00:00,4537.66034182
113
+ 1100,2.16133792425,0.144085657597,,0.184660971165,0.0001,0.150382646453,17231870,0.966789248586,,2026-07-14T20:00:28.615763+00:00,4577.17880887
114
+ 1110,2.18101328087,0.138406217098,,0.156137287617,0.0001,0.145552860852,17390150,0.968244227767,,2026-07-14T20:01:08.932935+00:00,4617.49598119
115
+ 1120,2.20068863748,0.149384868145,,0.188025072217,0.0001,0.158229852095,17547602,0.966268263757,,2026-07-14T20:01:48.411387+00:00,4656.97443159
116
+ 1130,2.2203639941,0.155771589279,,0.181949421763,0.0001,0.159846979193,17704171,0.964384664595,,2026-07-14T20:02:28.405767+00:00,4696.96881314
117
+ 1140,2.24003935071,0.150599765778,,0.185370415449,0.0001,0.159115260746,17861739,0.965643018484,,2026-07-14T20:03:07.586113+00:00,4736.14915861
118
+ 1150,2.25971470733,0.143140876293,,0.200799316168,0.0001,0.150902107451,18019056,0.966995790601,,2026-07-14T20:03:47.158333+00:00,4775.72137882
119
+ 1160,2.27939006394,0.150355446339,,0.182861015201,0.0001,0.15370692974,18175722,0.965399813652,,2026-07-14T20:04:26.406353+00:00,4814.96939667
120
+ 1170,2.29906542056,0.15268086195,,0.213103458285,0.0001,0.160970865935,18331478,0.965237861872,,2026-07-14T20:05:06.693820+00:00,4855.25686449
121
+ 1180,2.31874077718,0.14596208334,,0.158253356814,0.0001,0.151732229069,18487987,0.966772224009,,2026-07-14T20:05:46.624121+00:00,4895.18716697
122
+ 1190,2.33841613379,0.144477832317,,0.146822363138,0.0001,0.148501190729,18645678,0.967359808087,,2026-07-14T20:06:27.198857+00:00,4935.76190269
123
+ 1200,2.35809149041,0.161581110954,,0.169999018312,0.0001,0.16736223828,18803500,0.963338895142,,2026-07-14T20:07:06.846514+00:00,4975.40955758
124
+ 1210,2.37776684702,0.153406262398,,0.205819413066,0.0001,0.160941473302,18961366,0.965424561501,,2026-07-14T20:07:47.214742+00:00,5015.77778801
125
+ 1220,2.39744220364,0.136192810535,,0.190873071551,0.0001,0.142981468607,19119178,0.96834358573,,2026-07-14T20:08:26.742055+00:00,5055.30510108
126
+ 1230,2.41711756026,0.141453588009,,0.159980386496,0.0001,0.145217145327,19275993,0.968073041737,,2026-07-14T20:09:06.961714+00:00,5095.52476001
127
+ 1240,2.43679291687,0.148014056683,,0.153729602695,0.0001,0.158173099998,19432551,0.966497783363,,2026-07-14T20:09:46.420724+00:00,5134.98376955
128
+ 1250,2.45646827349,0.163712227345,,0.168259471655,0.0001,0.167730315216,19589038,0.962808711827,,2026-07-14T20:10:26.663478+00:00,5175.22652365
129
+ 1260,2.4761436301,0.127088427544,,0.165303364396,0.0001,0.133041086979,19745335,0.970423710346,,2026-07-14T20:11:05.915422+00:00,5214.47846581
130
+ 1270,2.49581898672,0.159169518948,,0.21293412149,0.0001,0.165325385705,19901135,0.963794569671,,2026-07-14T20:11:46.061351+00:00,5254.62439666
131
+ 1280,2.51549434333,0.154375171661,,0.197351962328,0.0001,0.159839222021,20058459,0.964553439617,,2026-07-14T20:12:25.361773+00:00,5293.9248188
132
+ 1290,2.53516969995,0.153488874435,,0.18740414083,0.0001,0.161631961539,20216596,0.964601299167,,2026-07-14T20:13:05.881831+00:00,5334.44487748
133
+ 1300,2.55484505657,0.147537398338,,0.178495824337,0.0001,0.154408513475,20374020,0.966325297952,,2026-07-14T20:13:45.465191+00:00,5374.02823627
134
+ 1310,2.57452041318,0.148759007454,,0.165089562535,0.0001,0.151681568567,20531138,0.966131106019,,2026-07-14T20:14:25.818427+00:00,5414.38147254
135
+ 1320,2.5941957698,0.168530833721,,0.183519527316,0.0001,0.176336678024,20686701,0.961573958397,,2026-07-14T20:15:04.997962+00:00,5453.56100702
136
+ 1330,2.61387112641,0.142667996883,,0.177817106247,0.0001,0.150646794215,20843865,0.967578789592,,2026-07-14T20:15:45.484718+00:00,5494.04776416
137
+ 1340,2.63354648303,0.168079650402,,0.189123302698,0.0001,0.17436240688,21000133,0.962020015717,,2026-07-14T20:16:24.643629+00:00,5533.20667599
138
+ 1350,2.65322183965,0.160590958595,,0.178365811706,0.0001,0.164238455892,21155516,0.964021879435,,2026-07-14T20:17:04.731592+00:00,5573.29463823
139
+ 1360,2.67289719626,0.165736973286,,0.17040604353,0.0001,0.173429191206,21311602,0.962552376091,,2026-07-14T20:17:44.004758+00:00,5612.56780189
140
+ 1370,2.69257255288,0.141857481003,,0.173059791327,0.0001,0.146332843136,21469833,0.967443062365,,2026-07-14T20:18:24.917303+00:00,5653.48034912
141
+ 1380,2.71224790949,0.144333839417,,0.15543037653,0.0001,0.150063864421,21626372,0.967242586613,,2026-07-14T20:19:03.713962+00:00,5692.27700782
142
+ 1390,2.73192326611,0.149240171909,,0.209980487823,0.0001,0.155178779084,21783541,0.965631118417,,2026-07-14T20:19:44.178706+00:00,5732.74175155
143
+ 1400,2.75159862273,0.140009224415,,0.170756846666,0.0001,0.14718689071,21941737,0.967410634458,,2026-07-14T20:20:24.400090+00:00,5772.96313486
144
+ 1410,2.77127397934,0.158368110657,,0.160231694579,0.0001,0.163548708614,22098896,0.963823059201,,2026-07-14T20:21:04.821885+00:00,5813.38493038
145
+ 1420,2.79094933596,0.15230704546,,0.16520537436,0.0001,0.156925874669,22253937,0.965254628658,,2026-07-14T20:21:43.554602+00:00,5852.1176477
146
+ 1430,2.81062469257,0.149685204029,,0.15979103744,0.0001,0.15920828823,22407704,0.965691076219,,2026-07-14T20:22:22.572894+00:00,5891.13593973
147
+ 1440,2.83030004919,0.141947734356,,0.17102535069,0.0001,0.143590484187,22566364,0.967758500576,,2026-07-14T20:23:01.967077+00:00,5930.53012291
148
+ 1450,2.8499754058,0.149233210087,,0.176346048713,0.0001,0.154821686074,22723504,0.965960656106,,2026-07-14T20:23:42.259042+00:00,5970.82208756
149
+ 1460,2.86965076242,0.149314463139,,0.167141124606,0.0001,0.157680661697,22881293,0.965893740952,,2026-07-14T20:24:21.692911+00:00,6010.25595651
150
+ 1470,2.88932611904,0.153538620472,,0.187396898866,0.0001,0.15704208063,23038652,0.964700344205,,2026-07-14T20:25:01.684654+00:00,6050.24770006
151
+ 1480,2.90900147565,0.146680307388,,0.161297917366,0.0001,0.154165752605,23195931,0.966039390862,,2026-07-14T20:25:41.291970+00:00,6089.85501546
152
+ 1490,2.92867683227,0.149410581589,,0.180445641279,0.0001,0.153569581266,23353688,0.965923695266,,2026-07-14T20:26:21.664709+00:00,6130.22775456
153
+ 1500,2.94835218888,0.149590086937,,0.183879300952,0.0001,0.155868681241,23510145,0.965885919333,,2026-07-14T20:27:00.725130+00:00,6169.28817651
154
+ 1510,2.9680275455,0.162458205223,,0.158856883645,0.0001,0.166671768948,23666330,0.963560299575,,2026-07-14T20:27:40.733926+00:00,6209.29697027
155
+ 1520,2.98770290212,0.161911118031,,0.233410969377,0.0001,0.170425816625,23825027,0.963155913353,,2026-07-14T20:28:20.899434+00:00,6249.46247936
156
+ 1527,3,0.202153051844,0.508632481098,,,,,,6374.1215,2026-07-14T20:30:25.555191+00:00,6374.11823728
training_history.jsonl ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"step": "10", "epoch": "0.0196753566158", "loss": "2.21839561462", "eval_loss": "", "grad_norm": "0.997185707092", "learning_rate": "0.0001", "entropy": "1.51200933158", "num_tokens": "158517", "mean_token_accuracy": "0.617804196477", "train_runtime": "", "timestamp": "2026-07-14T18:44:51.763785+00:00", "elapsed_seconds": "40.3268301521"}
2
+ {"step": "20", "epoch": "0.0393507132317", "loss": "1.0074344635", "eval_loss": "", "grad_norm": "1.03053867817", "learning_rate": "0.0001", "entropy": "1.0494144395", "num_tokens": "315386", "mean_token_accuracy": "0.796838776767", "train_runtime": "", "timestamp": "2026-07-14T18:45:31.035504+00:00", "elapsed_seconds": "79.598546861"}
3
+ {"step": "30", "epoch": "0.0590260698475", "loss": "0.544442272186", "eval_loss": "", "grad_norm": "0.422507971525", "learning_rate": "0.0001", "entropy": "0.548911824077", "num_tokens": "472899", "mean_token_accuracy": "0.898587629199", "train_runtime": "", "timestamp": "2026-07-14T18:46:11.591751+00:00", "elapsed_seconds": "120.154795361"}
4
+ {"step": "40", "epoch": "0.0787014264634", "loss": "0.445441532135", "eval_loss": "", "grad_norm": "0.18211427331", "learning_rate": "0.0001", "entropy": "0.455530235171", "num_tokens": "629242", "mean_token_accuracy": "0.916029234231", "train_runtime": "", "timestamp": "2026-07-14T18:46:50.703015+00:00", "elapsed_seconds": "159.266057246"}
5
+ {"step": "50", "epoch": "0.0983767830792", "loss": "0.451950931549", "eval_loss": "", "grad_norm": "0.175900384784", "learning_rate": "0.0001", "entropy": "0.460615222156", "num_tokens": "785551", "mean_token_accuracy": "0.911731639504", "train_runtime": "", "timestamp": "2026-07-14T18:47:30.680216+00:00", "elapsed_seconds": "199.243261393"}
6
+ {"step": "60", "epoch": "0.118052139695", "loss": "0.427107477188", "eval_loss": "", "grad_norm": "0.218168973923", "learning_rate": "0.0001", "entropy": "0.443916998804", "num_tokens": "942530", "mean_token_accuracy": "0.914710514247", "train_runtime": "", "timestamp": "2026-07-14T18:48:10.859048+00:00", "elapsed_seconds": "239.422092887"}
7
+ {"step": "70", "epoch": "0.137727496311", "loss": "0.397407078743", "eval_loss": "", "grad_norm": "0.257529675961", "learning_rate": "0.0001", "entropy": "0.416817698628", "num_tokens": "1100964", "mean_token_accuracy": "0.920619755983", "train_runtime": "", "timestamp": "2026-07-14T18:48:51.257122+00:00", "elapsed_seconds": "279.820165421"}
8
+ {"step": "80", "epoch": "0.157402852927", "loss": "0.35759768486", "eval_loss": "", "grad_norm": "0.660029649734", "learning_rate": "0.0001", "entropy": "0.384742297977", "num_tokens": "1257044", "mean_token_accuracy": "0.924064183235", "train_runtime": "", "timestamp": "2026-07-14T18:49:30.413227+00:00", "elapsed_seconds": "318.97627022"}
9
+ {"step": "90", "epoch": "0.177078209543", "loss": "0.35855255127", "eval_loss": "", "grad_norm": "0.264365077019", "learning_rate": "0.0001", "entropy": "0.391139079258", "num_tokens": "1414621", "mean_token_accuracy": "0.923739472032", "train_runtime": "", "timestamp": "2026-07-14T18:50:10.999452+00:00", "elapsed_seconds": "359.562497077"}
10
+ {"step": "100", "epoch": "0.196753566158", "loss": "0.317103385925", "eval_loss": "", "grad_norm": "0.273194640875", "learning_rate": "0.0001", "entropy": "0.336241504923", "num_tokens": "1571856", "mean_token_accuracy": "0.932000268996", "train_runtime": "", "timestamp": "2026-07-14T18:50:50.553821+00:00", "elapsed_seconds": "399.116867291"}
11
+ {"step": "110", "epoch": "0.216428922774", "loss": "0.315398597717", "eval_loss": "", "grad_norm": "0.224106535316", "learning_rate": "0.0001", "entropy": "0.332903936133", "num_tokens": "1729714", "mean_token_accuracy": "0.933503301442", "train_runtime": "", "timestamp": "2026-07-14T18:51:31.107001+00:00", "elapsed_seconds": "439.670047054"}
12
+ {"step": "120", "epoch": "0.23610427939", "loss": "0.283270096779", "eval_loss": "", "grad_norm": "0.243651434779", "learning_rate": "0.0001", "entropy": "0.303971963003", "num_tokens": "1886650", "mean_token_accuracy": "0.938405682147", "train_runtime": "", "timestamp": "2026-07-14T18:52:10.758208+00:00", "elapsed_seconds": "479.321254092"}
13
+ {"step": "130", "epoch": "0.255779636006", "loss": "0.277483892441", "eval_loss": "", "grad_norm": "0.242042064667", "learning_rate": "0.0001", "entropy": "0.299890926853", "num_tokens": "2042273", "mean_token_accuracy": "0.940673132241", "train_runtime": "", "timestamp": "2026-07-14T18:52:50.714737+00:00", "elapsed_seconds": "519.277783558"}
14
+ {"step": "140", "epoch": "0.275454992622", "loss": "0.265112280846", "eval_loss": "", "grad_norm": "0.265905737877", "learning_rate": "0.0001", "entropy": "0.288989692181", "num_tokens": "2198932", "mean_token_accuracy": "0.942814520001", "train_runtime": "", "timestamp": "2026-07-14T18:53:30.145592+00:00", "elapsed_seconds": "558.708636761"}
15
+ {"step": "150", "epoch": "0.295130349238", "loss": "0.254311847687", "eval_loss": "", "grad_norm": "0.307109236717", "learning_rate": "0.0001", "entropy": "0.265480846912", "num_tokens": "2355875", "mean_token_accuracy": "0.945721656084", "train_runtime": "", "timestamp": "2026-07-14T18:54:10.566345+00:00", "elapsed_seconds": "599.12936811"}
16
+ {"step": "160", "epoch": "0.314805705853", "loss": "0.23861758709", "eval_loss": "", "grad_norm": "0.344656556845", "learning_rate": "0.0001", "entropy": "0.259159233049", "num_tokens": "2512427", "mean_token_accuracy": "0.948875762522", "train_runtime": "", "timestamp": "2026-07-14T18:54:50.199052+00:00", "elapsed_seconds": "638.762097128"}
17
+ {"step": "170", "epoch": "0.334481062469", "loss": "0.230272626877", "eval_loss": "", "grad_norm": "0.261819094419", "learning_rate": "0.0001", "entropy": "0.242634578422", "num_tokens": "2668770", "mean_token_accuracy": "0.950659604371", "train_runtime": "", "timestamp": "2026-07-14T18:55:30.244558+00:00", "elapsed_seconds": "678.807603146"}
18
+ {"step": "180", "epoch": "0.354156419085", "loss": "0.214873051643", "eval_loss": "", "grad_norm": "0.348735779524", "learning_rate": "0.0001", "entropy": "0.232795054093", "num_tokens": "2825377", "mean_token_accuracy": "0.954010221362", "train_runtime": "", "timestamp": "2026-07-14T18:56:09.792053+00:00", "elapsed_seconds": "718.355098992"}
19
+ {"step": "190", "epoch": "0.373831775701", "loss": "0.206166386604", "eval_loss": "", "grad_norm": "0.238673120737", "learning_rate": "0.0001", "entropy": "0.220631313324", "num_tokens": "2982619", "mean_token_accuracy": "0.955823786557", "train_runtime": "", "timestamp": "2026-07-14T18:56:50.404094+00:00", "elapsed_seconds": "758.96713975"}
20
+ {"step": "200", "epoch": "0.393507132317", "loss": "0.214299988747", "eval_loss": "", "grad_norm": "0.259331971407", "learning_rate": "0.0001", "entropy": "0.234532970004", "num_tokens": "3138939", "mean_token_accuracy": "0.954085548222", "train_runtime": "", "timestamp": "2026-07-14T18:57:29.415292+00:00", "elapsed_seconds": "797.978338393"}
21
+ {"step": "210", "epoch": "0.413182488933", "loss": "0.199846196175", "eval_loss": "", "grad_norm": "0.242847457528", "learning_rate": "0.0001", "entropy": "0.212221397832", "num_tokens": "3294561", "mean_token_accuracy": "0.95643684268", "train_runtime": "", "timestamp": "2026-07-14T18:58:09.741501+00:00", "elapsed_seconds": "838.304546847"}
22
+ {"step": "220", "epoch": "0.432857845548", "loss": "0.204166269302", "eval_loss": "", "grad_norm": "0.306981146336", "learning_rate": "0.0001", "entropy": "0.214753072709", "num_tokens": "3449991", "mean_token_accuracy": "0.956322589517", "train_runtime": "", "timestamp": "2026-07-14T18:58:49.004264+00:00", "elapsed_seconds": "877.567307537"}
23
+ {"step": "230", "epoch": "0.452533202164", "loss": "0.201543879509", "eval_loss": "", "grad_norm": "0.249639883637", "learning_rate": "0.0001", "entropy": "0.216364040598", "num_tokens": "3608785", "mean_token_accuracy": "0.956451144814", "train_runtime": "", "timestamp": "2026-07-14T18:59:29.554834+00:00", "elapsed_seconds": "918.117879928"}
24
+ {"step": "240", "epoch": "0.47220855878", "loss": "0.215132331848", "eval_loss": "", "grad_norm": "0.24389693141", "learning_rate": "0.0001", "entropy": "0.223825023696", "num_tokens": "3766845", "mean_token_accuracy": "0.953913928568", "train_runtime": "", "timestamp": "2026-07-14T19:00:09.144558+00:00", "elapsed_seconds": "957.707604611"}
25
+ {"step": "250", "epoch": "0.491883915396", "loss": "0.200074696541", "eval_loss": "", "grad_norm": "0.322559267282", "learning_rate": "0.0001", "entropy": "0.216750285402", "num_tokens": "3923032", "mean_token_accuracy": "0.957909616828", "train_runtime": "", "timestamp": "2026-07-14T19:00:49.451045+00:00", "elapsed_seconds": "998.014089783"}
26
+ {"step": "260", "epoch": "0.511559272012", "loss": "0.198160004616", "eval_loss": "", "grad_norm": "0.223266273737", "learning_rate": "0.0001", "entropy": "0.210631896928", "num_tokens": "4080458", "mean_token_accuracy": "0.957292881608", "train_runtime": "", "timestamp": "2026-07-14T19:01:28.755248+00:00", "elapsed_seconds": "1037.31829346"}
27
+ {"step": "270", "epoch": "0.531234628628", "loss": "0.206540870667", "eval_loss": "", "grad_norm": "0.225164785981", "learning_rate": "0.0001", "entropy": "0.218400148116", "num_tokens": "4236536", "mean_token_accuracy": "0.955662851036", "train_runtime": "", "timestamp": "2026-07-14T19:02:09.002715+00:00", "elapsed_seconds": "1077.56576105"}
28
+ {"step": "280", "epoch": "0.550909985243", "loss": "0.203662276268", "eval_loss": "", "grad_norm": "0.211283221841", "learning_rate": "0.0001", "entropy": "0.212336610816", "num_tokens": "4391657", "mean_token_accuracy": "0.955616168678", "train_runtime": "", "timestamp": "2026-07-14T19:02:48.202244+00:00", "elapsed_seconds": "1116.7652876"}
29
+ {"step": "290", "epoch": "0.570585341859", "loss": "0.198586702347", "eval_loss": "", "grad_norm": "0.29632806778", "learning_rate": "0.0001", "entropy": "0.207531724125", "num_tokens": "4548276", "mean_token_accuracy": "0.957449166477", "train_runtime": "", "timestamp": "2026-07-14T19:03:28.562368+00:00", "elapsed_seconds": "1157.12541444"}
30
+ {"step": "300", "epoch": "0.590260698475", "loss": "0.1769338727", "eval_loss": "", "grad_norm": "0.200340017676", "learning_rate": "0.0001", "entropy": "0.189514789172", "num_tokens": "4706349", "mean_token_accuracy": "0.961175942421", "train_runtime": "", "timestamp": "2026-07-14T19:04:08.011571+00:00", "elapsed_seconds": "1196.57461494"}
31
+ {"step": "310", "epoch": "0.609936055091", "loss": "0.180926573277", "eval_loss": "", "grad_norm": "0.217008456588", "learning_rate": "0.0001", "entropy": "0.196929321997", "num_tokens": "4863585", "mean_token_accuracy": "0.96009849906", "train_runtime": "", "timestamp": "2026-07-14T19:04:48.498705+00:00", "elapsed_seconds": "1237.06175009"}
32
+ {"step": "320", "epoch": "0.629611411707", "loss": "0.180410706997", "eval_loss": "", "grad_norm": "0.236014738679", "learning_rate": "0.0001", "entropy": "0.18719286155", "num_tokens": "5022342", "mean_token_accuracy": "0.960626663268", "train_runtime": "", "timestamp": "2026-07-14T19:05:28.044393+00:00", "elapsed_seconds": "1276.60743859"}
33
+ {"step": "330", "epoch": "0.649286768323", "loss": "0.187188267708", "eval_loss": "", "grad_norm": "0.259057343006", "learning_rate": "0.0001", "entropy": "0.197457621247", "num_tokens": "5177117", "mean_token_accuracy": "0.958998320997", "train_runtime": "", "timestamp": "2026-07-14T19:06:07.805697+00:00", "elapsed_seconds": "1316.3687436"}
34
+ {"step": "340", "epoch": "0.668962124939", "loss": "0.203676509857", "eval_loss": "", "grad_norm": "0.230080202222", "learning_rate": "0.0001", "entropy": "0.207781772502", "num_tokens": "5334589", "mean_token_accuracy": "0.956454566121", "train_runtime": "", "timestamp": "2026-07-14T19:06:47.379224+00:00", "elapsed_seconds": "1355.94226928"}
35
+ {"step": "350", "epoch": "0.688637481554", "loss": "0.174643802643", "eval_loss": "", "grad_norm": "0.283878415823", "learning_rate": "0.0001", "entropy": "0.185111483", "num_tokens": "5491264", "mean_token_accuracy": "0.961374931037", "train_runtime": "", "timestamp": "2026-07-14T19:07:27.523734+00:00", "elapsed_seconds": "1396.08677869"}
36
+ {"step": "360", "epoch": "0.70831283817", "loss": "0.179219591618", "eval_loss": "", "grad_norm": "0.24345164001", "learning_rate": "0.0001", "entropy": "0.196479166858", "num_tokens": "5648766", "mean_token_accuracy": "0.96104246974", "train_runtime": "", "timestamp": "2026-07-14T19:08:07.102058+00:00", "elapsed_seconds": "1435.66510349"}
37
+ {"step": "370", "epoch": "0.727988194786", "loss": "0.172455787659", "eval_loss": "", "grad_norm": "0.217018336058", "learning_rate": "0.0001", "entropy": "0.181602787226", "num_tokens": "5805950", "mean_token_accuracy": "0.961948390305", "train_runtime": "", "timestamp": "2026-07-14T19:08:47.480806+00:00", "elapsed_seconds": "1476.04385117"}
38
+ {"step": "380", "epoch": "0.747663551402", "loss": "0.17482407093", "eval_loss": "", "grad_norm": "0.328721493483", "learning_rate": "0.0001", "entropy": "0.180027039908", "num_tokens": "5962530", "mean_token_accuracy": "0.961873859167", "train_runtime": "", "timestamp": "2026-07-14T19:09:26.926417+00:00", "elapsed_seconds": "1515.48946346"}
39
+ {"step": "390", "epoch": "0.767338908018", "loss": "0.162100684643", "eval_loss": "", "grad_norm": "0.200087547302", "learning_rate": "0.0001", "entropy": "0.176972093433", "num_tokens": "6119998", "mean_token_accuracy": "0.964109444618", "train_runtime": "", "timestamp": "2026-07-14T19:10:07.159783+00:00", "elapsed_seconds": "1555.72282812"}
40
+ {"step": "400", "epoch": "0.787014264634", "loss": "0.183642709255", "eval_loss": "", "grad_norm": "0.196165248752", "learning_rate": "0.0001", "entropy": "0.189363373816", "num_tokens": "6276791", "mean_token_accuracy": "0.960221962631", "train_runtime": "", "timestamp": "2026-07-14T19:10:46.801680+00:00", "elapsed_seconds": "1595.36472525"}
41
+ {"step": "410", "epoch": "0.806689621249", "loss": "0.171787154675", "eval_loss": "", "grad_norm": "0.245862558484", "learning_rate": "0.0001", "entropy": "0.182321485505", "num_tokens": "6433516", "mean_token_accuracy": "0.961653226614", "train_runtime": "", "timestamp": "2026-07-14T19:11:27.245900+00:00", "elapsed_seconds": "1635.80894558"}
42
+ {"step": "420", "epoch": "0.826364977865", "loss": "0.182455682755", "eval_loss": "", "grad_norm": "0.325130581856", "learning_rate": "0.0001", "entropy": "0.191438901238", "num_tokens": "6590564", "mean_token_accuracy": "0.959084931016", "train_runtime": "", "timestamp": "2026-07-14T19:12:06.417007+00:00", "elapsed_seconds": "1674.98005193"}
43
+ {"step": "430", "epoch": "0.846040334481", "loss": "0.169191372395", "eval_loss": "", "grad_norm": "0.207633405924", "learning_rate": "0.0001", "entropy": "0.178573282529", "num_tokens": "6747353", "mean_token_accuracy": "0.963288144767", "train_runtime": "", "timestamp": "2026-07-14T19:12:46.362587+00:00", "elapsed_seconds": "1714.92563345"}
44
+ {"step": "440", "epoch": "0.865715691097", "loss": "0.180155730247", "eval_loss": "", "grad_norm": "0.206554561853", "learning_rate": "0.0001", "entropy": "0.18681143187", "num_tokens": "6902907", "mean_token_accuracy": "0.960263897479", "train_runtime": "", "timestamp": "2026-07-14T19:13:25.393102+00:00", "elapsed_seconds": "1753.95614771"}
45
+ {"step": "450", "epoch": "0.885391047713", "loss": "0.171831595898", "eval_loss": "", "grad_norm": "0.30511072278", "learning_rate": "0.0001", "entropy": "0.181087859161", "num_tokens": "7059033", "mean_token_accuracy": "0.962434618175", "train_runtime": "", "timestamp": "2026-07-14T19:14:05.571981+00:00", "elapsed_seconds": "1794.13502563"}
46
+ {"step": "460", "epoch": "0.905066404329", "loss": "0.180643391609", "eval_loss": "", "grad_norm": "0.196387454867", "learning_rate": "0.0001", "entropy": "0.188525561802", "num_tokens": "7216633", "mean_token_accuracy": "0.959945641458", "train_runtime": "", "timestamp": "2026-07-14T19:14:44.847650+00:00", "elapsed_seconds": "1833.4106959"}
47
+ {"step": "470", "epoch": "0.924741760944", "loss": "0.193449866772", "eval_loss": "", "grad_norm": "0.19179905951", "learning_rate": "0.0001", "entropy": "0.196602920815", "num_tokens": "7373319", "mean_token_accuracy": "0.95812100023", "train_runtime": "", "timestamp": "2026-07-14T19:15:26.044097+00:00", "elapsed_seconds": "1874.60714292"}
48
+ {"step": "480", "epoch": "0.94441711756", "loss": "0.177313435078", "eval_loss": "", "grad_norm": "0.155064806342", "learning_rate": "0.0001", "entropy": "0.188440449163", "num_tokens": "7530283", "mean_token_accuracy": "0.961460134387", "train_runtime": "", "timestamp": "2026-07-14T19:16:06.338099+00:00", "elapsed_seconds": "1914.90114522"}
49
+ {"step": "490", "epoch": "0.964092474176", "loss": "0.180543804169", "eval_loss": "", "grad_norm": "0.233374238014", "learning_rate": "0.0001", "entropy": "0.187788728997", "num_tokens": "7687237", "mean_token_accuracy": "0.960137434304", "train_runtime": "", "timestamp": "2026-07-14T19:16:46.621608+00:00", "elapsed_seconds": "1955.18465457"}
50
+ {"step": "500", "epoch": "0.983767830792", "loss": "0.180359601974", "eval_loss": "", "grad_norm": "0.180976226926", "learning_rate": "0.0001", "entropy": "0.188501783088", "num_tokens": "7844660", "mean_token_accuracy": "0.96057779789", "train_runtime": "", "timestamp": "2026-07-14T19:17:25.826646+00:00", "elapsed_seconds": "1994.38969213"}
51
+ {"step": "509", "epoch": "1", "loss": "", "eval_loss": "0.480833500624", "grad_norm": "", "learning_rate": "", "entropy": "", "num_tokens": "", "mean_token_accuracy": "", "train_runtime": "", "timestamp": "2026-07-14T19:19:37.994260+00:00", "elapsed_seconds": "2126.55730501"}
52
+ {"step": "510", "epoch": "1.00196753566", "loss": "0.192923069", "eval_loss": "", "grad_norm": "0.169916674495", "learning_rate": "0.0001", "entropy": "0.198045271675", "num_tokens": "7988594", "mean_token_accuracy": "0.957474518467", "train_runtime": "", "timestamp": "2026-07-14T19:19:42.166818+00:00", "elapsed_seconds": "2130.72986391"}
53
+ {"step": "520", "epoch": "1.02164289228", "loss": "0.170230865479", "eval_loss": "", "grad_norm": "0.194753885269", "learning_rate": "0.0001", "entropy": "0.178488218039", "num_tokens": "8144390", "mean_token_accuracy": "0.961938048899", "train_runtime": "", "timestamp": "2026-07-14T19:20:21.674456+00:00", "elapsed_seconds": "2170.23750172"}
54
+ {"step": "530", "epoch": "1.04131824889", "loss": "0.156556117535", "eval_loss": "", "grad_norm": "0.168322607875", "learning_rate": "0.0001", "entropy": "0.165742422082", "num_tokens": "8300636", "mean_token_accuracy": "0.965336105227", "train_runtime": "", "timestamp": "2026-07-14T19:21:01.779702+00:00", "elapsed_seconds": "2210.3427483"}
55
+ {"step": "540", "epoch": "1.06099360551", "loss": "0.160328769684", "eval_loss": "", "grad_norm": "0.186233431101", "learning_rate": "0.0001", "entropy": "0.168560723402", "num_tokens": "8458859", "mean_token_accuracy": "0.96414308995", "train_runtime": "", "timestamp": "2026-07-14T19:21:41.348379+00:00", "elapsed_seconds": "2249.91142491"}
56
+ {"step": "550", "epoch": "1.08066896212", "loss": "0.169278275967", "eval_loss": "", "grad_norm": "0.182504862547", "learning_rate": "0.0001", "entropy": "0.173748625629", "num_tokens": "8615963", "mean_token_accuracy": "0.963006226718", "train_runtime": "", "timestamp": "2026-07-14T19:22:21.535345+00:00", "elapsed_seconds": "2290.09839124"}
57
+ {"step": "560", "epoch": "1.10034431874", "loss": "0.15963177681", "eval_loss": "", "grad_norm": "0.230876639485", "learning_rate": "0.0001", "entropy": "0.170929611102", "num_tokens": "8772742", "mean_token_accuracy": "0.964167003334", "train_runtime": "", "timestamp": "2026-07-14T19:23:01.254806+00:00", "elapsed_seconds": "2329.81785137"}
58
+ {"step": "570", "epoch": "1.12001967536", "loss": "0.157126617432", "eval_loss": "", "grad_norm": "0.179838463664", "learning_rate": "0.0001", "entropy": "0.16031345129", "num_tokens": "8930284", "mean_token_accuracy": "0.964271454513", "train_runtime": "", "timestamp": "2026-07-14T19:23:41.828440+00:00", "elapsed_seconds": "2370.3914869"}
59
+ {"step": "580", "epoch": "1.13969503197", "loss": "0.166155254841", "eval_loss": "", "grad_norm": "0.223575741053", "learning_rate": "0.0001", "entropy": "0.175235996023", "num_tokens": "9088333", "mean_token_accuracy": "0.963411171734", "train_runtime": "", "timestamp": "2026-07-14T19:24:21.736483+00:00", "elapsed_seconds": "2410.29952893"}
60
+ {"step": "590", "epoch": "1.15937038859", "loss": "0.157854199409", "eval_loss": "", "grad_norm": "0.207558766007", "learning_rate": "0.0001", "entropy": "0.166331799515", "num_tokens": "9244311", "mean_token_accuracy": "0.964486956596", "train_runtime": "", "timestamp": "2026-07-14T19:25:02.048624+00:00", "elapsed_seconds": "2450.61167042"}
61
+ {"step": "600", "epoch": "1.1790457452", "loss": "0.162935483456", "eval_loss": "", "grad_norm": "0.173033863306", "learning_rate": "0.0001", "entropy": "0.167879271507", "num_tokens": "9402415", "mean_token_accuracy": "0.963171319664", "train_runtime": "", "timestamp": "2026-07-14T19:25:42.017999+00:00", "elapsed_seconds": "2490.58104435"}
62
+ {"step": "610", "epoch": "1.19872110182", "loss": "0.17940980196", "eval_loss": "", "grad_norm": "0.200073346496", "learning_rate": "0.0001", "entropy": "0.188126351219", "num_tokens": "9556833", "mean_token_accuracy": "0.960648794472", "train_runtime": "", "timestamp": "2026-07-14T19:26:22.097129+00:00", "elapsed_seconds": "2530.6601727"}
63
+ {"step": "620", "epoch": "1.21839645844", "loss": "0.165559697151", "eval_loss": "", "grad_norm": "0.176193639636", "learning_rate": "0.0001", "entropy": "0.171528098173", "num_tokens": "9714854", "mean_token_accuracy": "0.962894457579", "train_runtime": "", "timestamp": "2026-07-14T19:27:01.449098+00:00", "elapsed_seconds": "2570.01214384"}
64
+ {"step": "630", "epoch": "1.23807181505", "loss": "0.192057442665", "eval_loss": "", "grad_norm": "0.249208107591", "learning_rate": "0.0001", "entropy": "0.198998957127", "num_tokens": "9871371", "mean_token_accuracy": "0.957532766461", "train_runtime": "", "timestamp": "2026-07-14T19:27:41.929207+00:00", "elapsed_seconds": "2610.49225355"}
65
+ {"step": "640", "epoch": "1.25774717167", "loss": "0.16299200058", "eval_loss": "", "grad_norm": "0.19560906291", "learning_rate": "0.0001", "entropy": "0.172278838418", "num_tokens": "10028132", "mean_token_accuracy": "0.964198562503", "train_runtime": "", "timestamp": "2026-07-14T19:28:21.258453+00:00", "elapsed_seconds": "2649.82149892"}
66
+ {"step": "650", "epoch": "1.27742252828", "loss": "0.182843160629", "eval_loss": "", "grad_norm": "0.250791400671", "learning_rate": "0.0001", "entropy": "0.185129570402", "num_tokens": "10183998", "mean_token_accuracy": "0.959990593791", "train_runtime": "", "timestamp": "2026-07-14T19:29:01.072607+00:00", "elapsed_seconds": "2689.6356517"}
67
+ {"step": "660", "epoch": "1.2970978849", "loss": "0.156658506393", "eval_loss": "", "grad_norm": "0.173409461975", "learning_rate": "0.0001", "entropy": "0.165276159532", "num_tokens": "10338966", "mean_token_accuracy": "0.96491754353", "train_runtime": "", "timestamp": "2026-07-14T19:29:39.981675+00:00", "elapsed_seconds": "2728.5447211"}
68
+ {"step": "670", "epoch": "1.31677324152", "loss": "0.156918954849", "eval_loss": "", "grad_norm": "0.2154648453", "learning_rate": "0.0001", "entropy": "0.165447768662", "num_tokens": "10494154", "mean_token_accuracy": "0.964841979742", "train_runtime": "", "timestamp": "2026-07-14T19:30:20.537538+00:00", "elapsed_seconds": "2769.10058474"}
69
+ {"step": "680", "epoch": "1.33644859813", "loss": "0.168604540825", "eval_loss": "", "grad_norm": "0.202249929309", "learning_rate": "0.0001", "entropy": "0.174681568705", "num_tokens": "10650728", "mean_token_accuracy": "0.962572240829", "train_runtime": "", "timestamp": "2026-07-14T19:30:59.771538+00:00", "elapsed_seconds": "2808.33458337"}
70
+ {"step": "690", "epoch": "1.35612395475", "loss": "0.179817044735", "eval_loss": "", "grad_norm": "0.180585950613", "learning_rate": "0.0001", "entropy": "0.184981219843", "num_tokens": "10808084", "mean_token_accuracy": "0.960407233238", "train_runtime": "", "timestamp": "2026-07-14T19:31:40.333487+00:00", "elapsed_seconds": "2848.89653316"}
71
+ {"step": "700", "epoch": "1.37579931136", "loss": "0.17117511034", "eval_loss": "", "grad_norm": "0.189327448606", "learning_rate": "0.0001", "entropy": "0.179342577234", "num_tokens": "10965357", "mean_token_accuracy": "0.962315896153", "train_runtime": "", "timestamp": "2026-07-14T19:32:20.034892+00:00", "elapsed_seconds": "2888.597938"}
72
+ {"step": "710", "epoch": "1.39547466798", "loss": "0.163115501404", "eval_loss": "", "grad_norm": "0.14550216496", "learning_rate": "0.0001", "entropy": "0.170794568304", "num_tokens": "11122970", "mean_token_accuracy": "0.963731876016", "train_runtime": "", "timestamp": "2026-07-14T19:33:00.055825+00:00", "elapsed_seconds": "2928.61887098"}
73
+ {"step": "720", "epoch": "1.41515002459", "loss": "0.166223096848", "eval_loss": "", "grad_norm": "0.197181358933", "learning_rate": "0.0001", "entropy": "0.174024736509", "num_tokens": "11279748", "mean_token_accuracy": "0.962751935422", "train_runtime": "", "timestamp": "2026-07-14T19:33:39.118755+00:00", "elapsed_seconds": "2967.68180075"}
74
+ {"step": "730", "epoch": "1.43482538121", "loss": "0.171106719971", "eval_loss": "", "grad_norm": "0.232188314199", "learning_rate": "0.0001", "entropy": "0.175083556771", "num_tokens": "11437649", "mean_token_accuracy": "0.962013469636", "train_runtime": "", "timestamp": "2026-07-14T19:34:19.410048+00:00", "elapsed_seconds": "3007.9730946"}
75
+ {"step": "740", "epoch": "1.45450073783", "loss": "0.159269690514", "eval_loss": "", "grad_norm": "0.179410859942", "learning_rate": "0.0001", "entropy": "0.167735586967", "num_tokens": "11595537", "mean_token_accuracy": "0.964913637936", "train_runtime": "", "timestamp": "2026-07-14T19:34:59.021640+00:00", "elapsed_seconds": "3047.58468635"}
76
+ {"step": "750", "epoch": "1.47417609444", "loss": "0.168200385571", "eval_loss": "", "grad_norm": "0.155719816685", "learning_rate": "0.0001", "entropy": "0.175214201398", "num_tokens": "11753054", "mean_token_accuracy": "0.962682126462", "train_runtime": "", "timestamp": "2026-07-14T19:35:39.483000+00:00", "elapsed_seconds": "3088.04604582"}
77
+ {"step": "760", "epoch": "1.49385145106", "loss": "0.14346704483", "eval_loss": "", "grad_norm": "0.221360266209", "learning_rate": "0.0001", "entropy": "0.154037842248", "num_tokens": "11911753", "mean_token_accuracy": "0.968222369254", "train_runtime": "", "timestamp": "2026-07-14T19:36:18.999097+00:00", "elapsed_seconds": "3127.56214216"}
78
+ {"step": "770", "epoch": "1.51352680767", "loss": "0.161464035511", "eval_loss": "", "grad_norm": "0.167245700955", "learning_rate": "0.0001", "entropy": "0.165353834908", "num_tokens": "12069142", "mean_token_accuracy": "0.963533814251", "train_runtime": "", "timestamp": "2026-07-14T19:36:59.202932+00:00", "elapsed_seconds": "3167.76597772"}
79
+ {"step": "780", "epoch": "1.53320216429", "loss": "0.156456899643", "eval_loss": "", "grad_norm": "0.17038269341", "learning_rate": "0.0001", "entropy": "0.165936999302", "num_tokens": "12224271", "mean_token_accuracy": "0.96456181556", "train_runtime": "", "timestamp": "2026-07-14T19:37:37.739912+00:00", "elapsed_seconds": "3206.30295782"}
80
+ {"step": "790", "epoch": "1.55287752091", "loss": "0.162909126282", "eval_loss": "", "grad_norm": "0.194951727986", "learning_rate": "0.0001", "entropy": "0.166183103435", "num_tokens": "12380914", "mean_token_accuracy": "0.963615015149", "train_runtime": "", "timestamp": "2026-07-14T19:38:17.951802+00:00", "elapsed_seconds": "3246.51484807"}
81
+ {"step": "800", "epoch": "1.57255287752", "loss": "0.162587237358", "eval_loss": "", "grad_norm": "0.175505697727", "learning_rate": "0.0001", "entropy": "0.170294051804", "num_tokens": "12538265", "mean_token_accuracy": "0.963853444159", "train_runtime": "", "timestamp": "2026-07-14T19:38:57.359451+00:00", "elapsed_seconds": "3285.92249702"}
82
+ {"step": "810", "epoch": "1.59222823414", "loss": "0.16598303318", "eval_loss": "", "grad_norm": "0.182322755456", "learning_rate": "0.0001", "entropy": "0.171620354988", "num_tokens": "12694472", "mean_token_accuracy": "0.962860387564", "train_runtime": "", "timestamp": "2026-07-14T19:39:37.707953+00:00", "elapsed_seconds": "3326.2709987"}
83
+ {"step": "820", "epoch": "1.61190359075", "loss": "0.166602897644", "eval_loss": "", "grad_norm": "0.162591278553", "learning_rate": "0.0001", "entropy": "0.173711701855", "num_tokens": "12852241", "mean_token_accuracy": "0.962746998668", "train_runtime": "", "timestamp": "2026-07-14T19:40:17.543353+00:00", "elapsed_seconds": "3366.10638842"}
84
+ {"step": "830", "epoch": "1.63157894737", "loss": "0.173879754543", "eval_loss": "", "grad_norm": "0.179233163595", "learning_rate": "0.0001", "entropy": "0.180813023262", "num_tokens": "13010313", "mean_token_accuracy": "0.960821950436", "train_runtime": "", "timestamp": "2026-07-14T19:40:58.733570+00:00", "elapsed_seconds": "3407.29661541"}
85
+ {"step": "840", "epoch": "1.65125430398", "loss": "0.175928378105", "eval_loss": "", "grad_norm": "0.155069187284", "learning_rate": "0.0001", "entropy": "0.180705674551", "num_tokens": "13167919", "mean_token_accuracy": "0.960775198042", "train_runtime": "", "timestamp": "2026-07-14T19:41:38.127878+00:00", "elapsed_seconds": "3446.69092196"}
86
+ {"step": "850", "epoch": "1.6709296606", "loss": "0.156295228004", "eval_loss": "", "grad_norm": "0.187289491296", "learning_rate": "0.0001", "entropy": "0.163406656217", "num_tokens": "13326698", "mean_token_accuracy": "0.964838838577", "train_runtime": "", "timestamp": "2026-07-14T19:42:18.752868+00:00", "elapsed_seconds": "3487.3159135"}
87
+ {"step": "860", "epoch": "1.69060501722", "loss": "0.166846621037", "eval_loss": "", "grad_norm": "0.162943869829", "learning_rate": "0.0001", "entropy": "0.173019717634", "num_tokens": "13483225", "mean_token_accuracy": "0.962671551108", "train_runtime": "", "timestamp": "2026-07-14T19:42:57.972132+00:00", "elapsed_seconds": "3526.53517718"}
88
+ {"step": "870", "epoch": "1.71028037383", "loss": "0.175749349594", "eval_loss": "", "grad_norm": "0.174129873514", "learning_rate": "0.0001", "entropy": "0.181726143695", "num_tokens": "13639265", "mean_token_accuracy": "0.960963216424", "train_runtime": "", "timestamp": "2026-07-14T19:43:37.800757+00:00", "elapsed_seconds": "3566.36380294"}
89
+ {"step": "880", "epoch": "1.72995573045", "loss": "0.147835624218", "eval_loss": "", "grad_norm": "0.167158782482", "learning_rate": "0.0001", "entropy": "0.154400302656", "num_tokens": "13797228", "mean_token_accuracy": "0.966271243989", "train_runtime": "", "timestamp": "2026-07-14T19:44:17.466713+00:00", "elapsed_seconds": "3606.02975898"}
90
+ {"step": "890", "epoch": "1.74963108706", "loss": "0.16417195797", "eval_loss": "", "grad_norm": "0.196655958891", "learning_rate": "0.0001", "entropy": "0.170779196545", "num_tokens": "13954300", "mean_token_accuracy": "0.963016964495", "train_runtime": "", "timestamp": "2026-07-14T19:44:57.876500+00:00", "elapsed_seconds": "3646.43954604"}
91
+ {"step": "900", "epoch": "1.76930644368", "loss": "0.156272256374", "eval_loss": "", "grad_norm": "0.16130194068", "learning_rate": "0.0001", "entropy": "0.164454833046", "num_tokens": "14111213", "mean_token_accuracy": "0.964544859529", "train_runtime": "", "timestamp": "2026-07-14T19:45:37.195299+00:00", "elapsed_seconds": "3685.75834587"}
92
+ {"step": "910", "epoch": "1.7889818003", "loss": "0.163049316406", "eval_loss": "", "grad_norm": "0.136302247643", "learning_rate": "0.0001", "entropy": "0.169371592253", "num_tokens": "14267408", "mean_token_accuracy": "0.963653604686", "train_runtime": "", "timestamp": "2026-07-14T19:46:17.611788+00:00", "elapsed_seconds": "3726.17483362"}
93
+ {"step": "920", "epoch": "1.80865715691", "loss": "0.147612285614", "eval_loss": "", "grad_norm": "0.16987170279", "learning_rate": "0.0001", "entropy": "0.15369244311", "num_tokens": "14424482", "mean_token_accuracy": "0.966288872063", "train_runtime": "", "timestamp": "2026-07-14T19:46:57.069799+00:00", "elapsed_seconds": "3765.63284427"}
94
+ {"step": "930", "epoch": "1.82833251353", "loss": "0.151848042011", "eval_loss": "", "grad_norm": "0.155094489455", "learning_rate": "0.0001", "entropy": "0.158277668338", "num_tokens": "14581898", "mean_token_accuracy": "0.965862759948", "train_runtime": "", "timestamp": "2026-07-14T19:47:37.633555+00:00", "elapsed_seconds": "3806.19659956"}
95
+ {"step": "940", "epoch": "1.84800787014", "loss": "0.161747670174", "eval_loss": "", "grad_norm": "0.176779091358", "learning_rate": "0.0001", "entropy": "0.167998809647", "num_tokens": "14739682", "mean_token_accuracy": "0.963628257811", "train_runtime": "", "timestamp": "2026-07-14T19:48:16.925203+00:00", "elapsed_seconds": "3845.48824898"}
96
+ {"step": "950", "epoch": "1.86768322676", "loss": "0.175212359428", "eval_loss": "", "grad_norm": "0.186388149858", "learning_rate": "0.0001", "entropy": "0.179522324633", "num_tokens": "14897444", "mean_token_accuracy": "0.961017121375", "train_runtime": "", "timestamp": "2026-07-14T19:48:57.156443+00:00", "elapsed_seconds": "3885.71948934"}
97
+ {"step": "960", "epoch": "1.88735858337", "loss": "0.158414626122", "eval_loss": "", "grad_norm": "0.143919214606", "learning_rate": "0.0001", "entropy": "0.163319427148", "num_tokens": "15053687", "mean_token_accuracy": "0.964056003094", "train_runtime": "", "timestamp": "2026-07-14T19:49:36.454119+00:00", "elapsed_seconds": "3925.01716572"}
98
+ {"step": "970", "epoch": "1.90703393999", "loss": "0.176569092274", "eval_loss": "", "grad_norm": "0.212138965726", "learning_rate": "0.0001", "entropy": "0.18215863388", "num_tokens": "15210316", "mean_token_accuracy": "0.960795244575", "train_runtime": "", "timestamp": "2026-07-14T19:50:16.690489+00:00", "elapsed_seconds": "3965.25353464"}
99
+ {"step": "980", "epoch": "1.92670929661", "loss": "0.158397960663", "eval_loss": "", "grad_norm": "0.177053377032", "learning_rate": "0.0001", "entropy": "0.166588350572", "num_tokens": "15367289", "mean_token_accuracy": "0.964338231087", "train_runtime": "", "timestamp": "2026-07-14T19:50:55.920110+00:00", "elapsed_seconds": "4004.48315612"}
100
+ {"step": "990", "epoch": "1.94638465322", "loss": "0.172202885151", "eval_loss": "", "grad_norm": "0.198846206069", "learning_rate": "0.0001", "entropy": "0.173680017237", "num_tokens": "15519567", "mean_token_accuracy": "0.962071706355", "train_runtime": "", "timestamp": "2026-07-14T19:51:34.604574+00:00", "elapsed_seconds": "4043.16762026"}
101
+ {"step": "1000", "epoch": "1.96606000984", "loss": "0.135841083527", "eval_loss": "", "grad_norm": "0.173086538911", "learning_rate": "0.0001", "entropy": "0.145909713674", "num_tokens": "15677796", "mean_token_accuracy": "0.968235512078", "train_runtime": "", "timestamp": "2026-07-14T19:52:13.740978+00:00", "elapsed_seconds": "4082.30402441"}
102
+ {"step": "1010", "epoch": "1.98573536645", "loss": "0.169320774078", "eval_loss": "", "grad_norm": "0.182816982269", "learning_rate": "0.0001", "entropy": "0.173335280083", "num_tokens": "15834854", "mean_token_accuracy": "0.962027944624", "train_runtime": "", "timestamp": "2026-07-14T19:52:54.082399+00:00", "elapsed_seconds": "4122.64544415"}
103
+ {"step": "1018", "epoch": "2", "loss": "", "eval_loss": "0.503623247147", "grad_norm": "", "learning_rate": "", "entropy": "", "num_tokens": "", "mean_token_accuracy": "", "train_runtime": "", "timestamp": "2026-07-14T19:55:03.085112+00:00", "elapsed_seconds": "4251.64815671"}
104
+ {"step": "1020", "epoch": "2.00393507132", "loss": "0.156750667095", "eval_loss": "", "grad_norm": "0.21632951498", "learning_rate": "0.0001", "entropy": "0.174239883048", "num_tokens": "15980450", "mean_token_accuracy": "0.96212148022", "train_runtime": "", "timestamp": "2026-07-14T19:55:11.229820+00:00", "elapsed_seconds": "4259.79286485"}
105
+ {"step": "1030", "epoch": "2.02361042794", "loss": "0.16210129261", "eval_loss": "", "grad_norm": "0.189968839288", "learning_rate": "0.0001", "entropy": "0.16485402789", "num_tokens": "16135755", "mean_token_accuracy": "0.963425774872", "train_runtime": "", "timestamp": "2026-07-14T19:55:51.178359+00:00", "elapsed_seconds": "4299.74140436"}
106
+ {"step": "1040", "epoch": "2.04328578455", "loss": "0.151982796192", "eval_loss": "", "grad_norm": "0.186117559671", "learning_rate": "0.0001", "entropy": "0.164776344784", "num_tokens": "16293622", "mean_token_accuracy": "0.96450523138", "train_runtime": "", "timestamp": "2026-07-14T19:56:30.672001+00:00", "elapsed_seconds": "4339.23504656"}
107
+ {"step": "1050", "epoch": "2.06296114117", "loss": "0.163860476017", "eval_loss": "", "grad_norm": "0.191996067762", "learning_rate": "0.0001", "entropy": "0.16807874972", "num_tokens": "16449011", "mean_token_accuracy": "0.962708187103", "train_runtime": "", "timestamp": "2026-07-14T19:57:10.452525+00:00", "elapsed_seconds": "4379.01557092"}
108
+ {"step": "1060", "epoch": "2.08263649779", "loss": "0.172380006313", "eval_loss": "", "grad_norm": "0.209191724658", "learning_rate": "0.0001", "entropy": "0.179409135506", "num_tokens": "16606544", "mean_token_accuracy": "0.961136734486", "train_runtime": "", "timestamp": "2026-07-14T19:57:49.831844+00:00", "elapsed_seconds": "4418.39488819"}
109
+ {"step": "1070", "epoch": "2.1023118544", "loss": "0.141785228252", "eval_loss": "", "grad_norm": "0.14348590374", "learning_rate": "0.0001", "entropy": "0.150052259676", "num_tokens": "16762655", "mean_token_accuracy": "0.967867693305", "train_runtime": "", "timestamp": "2026-07-14T19:58:29.903173+00:00", "elapsed_seconds": "4458.46621657"}
110
+ {"step": "1080", "epoch": "2.12198721102", "loss": "0.158943665028", "eval_loss": "", "grad_norm": "0.173628717661", "learning_rate": "0.0001", "entropy": "0.165021989681", "num_tokens": "16918189", "mean_token_accuracy": "0.963940601051", "train_runtime": "", "timestamp": "2026-07-14T19:59:09.353150+00:00", "elapsed_seconds": "4497.9161958"}
111
+ {"step": "1090", "epoch": "2.14166256763", "loss": "0.161983144283", "eval_loss": "", "grad_norm": "0.170995950699", "learning_rate": "0.0001", "entropy": "0.169115608837", "num_tokens": "17074380", "mean_token_accuracy": "0.9631606251", "train_runtime": "", "timestamp": "2026-07-14T19:59:49.097296+00:00", "elapsed_seconds": "4537.66034182"}
112
+ {"step": "1100", "epoch": "2.16133792425", "loss": "0.144085657597", "eval_loss": "", "grad_norm": "0.184660971165", "learning_rate": "0.0001", "entropy": "0.150382646453", "num_tokens": "17231870", "mean_token_accuracy": "0.966789248586", "train_runtime": "", "timestamp": "2026-07-14T20:00:28.615763+00:00", "elapsed_seconds": "4577.17880887"}
113
+ {"step": "1110", "epoch": "2.18101328087", "loss": "0.138406217098", "eval_loss": "", "grad_norm": "0.156137287617", "learning_rate": "0.0001", "entropy": "0.145552860852", "num_tokens": "17390150", "mean_token_accuracy": "0.968244227767", "train_runtime": "", "timestamp": "2026-07-14T20:01:08.932935+00:00", "elapsed_seconds": "4617.49598119"}
114
+ {"step": "1120", "epoch": "2.20068863748", "loss": "0.149384868145", "eval_loss": "", "grad_norm": "0.188025072217", "learning_rate": "0.0001", "entropy": "0.158229852095", "num_tokens": "17547602", "mean_token_accuracy": "0.966268263757", "train_runtime": "", "timestamp": "2026-07-14T20:01:48.411387+00:00", "elapsed_seconds": "4656.97443159"}
115
+ {"step": "1130", "epoch": "2.2203639941", "loss": "0.155771589279", "eval_loss": "", "grad_norm": "0.181949421763", "learning_rate": "0.0001", "entropy": "0.159846979193", "num_tokens": "17704171", "mean_token_accuracy": "0.964384664595", "train_runtime": "", "timestamp": "2026-07-14T20:02:28.405767+00:00", "elapsed_seconds": "4696.96881314"}
116
+ {"step": "1140", "epoch": "2.24003935071", "loss": "0.150599765778", "eval_loss": "", "grad_norm": "0.185370415449", "learning_rate": "0.0001", "entropy": "0.159115260746", "num_tokens": "17861739", "mean_token_accuracy": "0.965643018484", "train_runtime": "", "timestamp": "2026-07-14T20:03:07.586113+00:00", "elapsed_seconds": "4736.14915861"}
117
+ {"step": "1150", "epoch": "2.25971470733", "loss": "0.143140876293", "eval_loss": "", "grad_norm": "0.200799316168", "learning_rate": "0.0001", "entropy": "0.150902107451", "num_tokens": "18019056", "mean_token_accuracy": "0.966995790601", "train_runtime": "", "timestamp": "2026-07-14T20:03:47.158333+00:00", "elapsed_seconds": "4775.72137882"}
118
+ {"step": "1160", "epoch": "2.27939006394", "loss": "0.150355446339", "eval_loss": "", "grad_norm": "0.182861015201", "learning_rate": "0.0001", "entropy": "0.15370692974", "num_tokens": "18175722", "mean_token_accuracy": "0.965399813652", "train_runtime": "", "timestamp": "2026-07-14T20:04:26.406353+00:00", "elapsed_seconds": "4814.96939667"}
119
+ {"step": "1170", "epoch": "2.29906542056", "loss": "0.15268086195", "eval_loss": "", "grad_norm": "0.213103458285", "learning_rate": "0.0001", "entropy": "0.160970865935", "num_tokens": "18331478", "mean_token_accuracy": "0.965237861872", "train_runtime": "", "timestamp": "2026-07-14T20:05:06.693820+00:00", "elapsed_seconds": "4855.25686449"}
120
+ {"step": "1180", "epoch": "2.31874077718", "loss": "0.14596208334", "eval_loss": "", "grad_norm": "0.158253356814", "learning_rate": "0.0001", "entropy": "0.151732229069", "num_tokens": "18487987", "mean_token_accuracy": "0.966772224009", "train_runtime": "", "timestamp": "2026-07-14T20:05:46.624121+00:00", "elapsed_seconds": "4895.18716697"}
121
+ {"step": "1190", "epoch": "2.33841613379", "loss": "0.144477832317", "eval_loss": "", "grad_norm": "0.146822363138", "learning_rate": "0.0001", "entropy": "0.148501190729", "num_tokens": "18645678", "mean_token_accuracy": "0.967359808087", "train_runtime": "", "timestamp": "2026-07-14T20:06:27.198857+00:00", "elapsed_seconds": "4935.76190269"}
122
+ {"step": "1200", "epoch": "2.35809149041", "loss": "0.161581110954", "eval_loss": "", "grad_norm": "0.169999018312", "learning_rate": "0.0001", "entropy": "0.16736223828", "num_tokens": "18803500", "mean_token_accuracy": "0.963338895142", "train_runtime": "", "timestamp": "2026-07-14T20:07:06.846514+00:00", "elapsed_seconds": "4975.40955758"}
123
+ {"step": "1210", "epoch": "2.37776684702", "loss": "0.153406262398", "eval_loss": "", "grad_norm": "0.205819413066", "learning_rate": "0.0001", "entropy": "0.160941473302", "num_tokens": "18961366", "mean_token_accuracy": "0.965424561501", "train_runtime": "", "timestamp": "2026-07-14T20:07:47.214742+00:00", "elapsed_seconds": "5015.77778801"}
124
+ {"step": "1220", "epoch": "2.39744220364", "loss": "0.136192810535", "eval_loss": "", "grad_norm": "0.190873071551", "learning_rate": "0.0001", "entropy": "0.142981468607", "num_tokens": "19119178", "mean_token_accuracy": "0.96834358573", "train_runtime": "", "timestamp": "2026-07-14T20:08:26.742055+00:00", "elapsed_seconds": "5055.30510108"}
125
+ {"step": "1230", "epoch": "2.41711756026", "loss": "0.141453588009", "eval_loss": "", "grad_norm": "0.159980386496", "learning_rate": "0.0001", "entropy": "0.145217145327", "num_tokens": "19275993", "mean_token_accuracy": "0.968073041737", "train_runtime": "", "timestamp": "2026-07-14T20:09:06.961714+00:00", "elapsed_seconds": "5095.52476001"}
126
+ {"step": "1240", "epoch": "2.43679291687", "loss": "0.148014056683", "eval_loss": "", "grad_norm": "0.153729602695", "learning_rate": "0.0001", "entropy": "0.158173099998", "num_tokens": "19432551", "mean_token_accuracy": "0.966497783363", "train_runtime": "", "timestamp": "2026-07-14T20:09:46.420724+00:00", "elapsed_seconds": "5134.98376955"}
127
+ {"step": "1250", "epoch": "2.45646827349", "loss": "0.163712227345", "eval_loss": "", "grad_norm": "0.168259471655", "learning_rate": "0.0001", "entropy": "0.167730315216", "num_tokens": "19589038", "mean_token_accuracy": "0.962808711827", "train_runtime": "", "timestamp": "2026-07-14T20:10:26.663478+00:00", "elapsed_seconds": "5175.22652365"}
128
+ {"step": "1260", "epoch": "2.4761436301", "loss": "0.127088427544", "eval_loss": "", "grad_norm": "0.165303364396", "learning_rate": "0.0001", "entropy": "0.133041086979", "num_tokens": "19745335", "mean_token_accuracy": "0.970423710346", "train_runtime": "", "timestamp": "2026-07-14T20:11:05.915422+00:00", "elapsed_seconds": "5214.47846581"}
129
+ {"step": "1270", "epoch": "2.49581898672", "loss": "0.159169518948", "eval_loss": "", "grad_norm": "0.21293412149", "learning_rate": "0.0001", "entropy": "0.165325385705", "num_tokens": "19901135", "mean_token_accuracy": "0.963794569671", "train_runtime": "", "timestamp": "2026-07-14T20:11:46.061351+00:00", "elapsed_seconds": "5254.62439666"}
130
+ {"step": "1280", "epoch": "2.51549434333", "loss": "0.154375171661", "eval_loss": "", "grad_norm": "0.197351962328", "learning_rate": "0.0001", "entropy": "0.159839222021", "num_tokens": "20058459", "mean_token_accuracy": "0.964553439617", "train_runtime": "", "timestamp": "2026-07-14T20:12:25.361773+00:00", "elapsed_seconds": "5293.9248188"}
131
+ {"step": "1290", "epoch": "2.53516969995", "loss": "0.153488874435", "eval_loss": "", "grad_norm": "0.18740414083", "learning_rate": "0.0001", "entropy": "0.161631961539", "num_tokens": "20216596", "mean_token_accuracy": "0.964601299167", "train_runtime": "", "timestamp": "2026-07-14T20:13:05.881831+00:00", "elapsed_seconds": "5334.44487748"}
132
+ {"step": "1300", "epoch": "2.55484505657", "loss": "0.147537398338", "eval_loss": "", "grad_norm": "0.178495824337", "learning_rate": "0.0001", "entropy": "0.154408513475", "num_tokens": "20374020", "mean_token_accuracy": "0.966325297952", "train_runtime": "", "timestamp": "2026-07-14T20:13:45.465191+00:00", "elapsed_seconds": "5374.02823627"}
133
+ {"step": "1310", "epoch": "2.57452041318", "loss": "0.148759007454", "eval_loss": "", "grad_norm": "0.165089562535", "learning_rate": "0.0001", "entropy": "0.151681568567", "num_tokens": "20531138", "mean_token_accuracy": "0.966131106019", "train_runtime": "", "timestamp": "2026-07-14T20:14:25.818427+00:00", "elapsed_seconds": "5414.38147254"}
134
+ {"step": "1320", "epoch": "2.5941957698", "loss": "0.168530833721", "eval_loss": "", "grad_norm": "0.183519527316", "learning_rate": "0.0001", "entropy": "0.176336678024", "num_tokens": "20686701", "mean_token_accuracy": "0.961573958397", "train_runtime": "", "timestamp": "2026-07-14T20:15:04.997962+00:00", "elapsed_seconds": "5453.56100702"}
135
+ {"step": "1330", "epoch": "2.61387112641", "loss": "0.142667996883", "eval_loss": "", "grad_norm": "0.177817106247", "learning_rate": "0.0001", "entropy": "0.150646794215", "num_tokens": "20843865", "mean_token_accuracy": "0.967578789592", "train_runtime": "", "timestamp": "2026-07-14T20:15:45.484718+00:00", "elapsed_seconds": "5494.04776416"}
136
+ {"step": "1340", "epoch": "2.63354648303", "loss": "0.168079650402", "eval_loss": "", "grad_norm": "0.189123302698", "learning_rate": "0.0001", "entropy": "0.17436240688", "num_tokens": "21000133", "mean_token_accuracy": "0.962020015717", "train_runtime": "", "timestamp": "2026-07-14T20:16:24.643629+00:00", "elapsed_seconds": "5533.20667599"}
137
+ {"step": "1350", "epoch": "2.65322183965", "loss": "0.160590958595", "eval_loss": "", "grad_norm": "0.178365811706", "learning_rate": "0.0001", "entropy": "0.164238455892", "num_tokens": "21155516", "mean_token_accuracy": "0.964021879435", "train_runtime": "", "timestamp": "2026-07-14T20:17:04.731592+00:00", "elapsed_seconds": "5573.29463823"}
138
+ {"step": "1360", "epoch": "2.67289719626", "loss": "0.165736973286", "eval_loss": "", "grad_norm": "0.17040604353", "learning_rate": "0.0001", "entropy": "0.173429191206", "num_tokens": "21311602", "mean_token_accuracy": "0.962552376091", "train_runtime": "", "timestamp": "2026-07-14T20:17:44.004758+00:00", "elapsed_seconds": "5612.56780189"}
139
+ {"step": "1370", "epoch": "2.69257255288", "loss": "0.141857481003", "eval_loss": "", "grad_norm": "0.173059791327", "learning_rate": "0.0001", "entropy": "0.146332843136", "num_tokens": "21469833", "mean_token_accuracy": "0.967443062365", "train_runtime": "", "timestamp": "2026-07-14T20:18:24.917303+00:00", "elapsed_seconds": "5653.48034912"}
140
+ {"step": "1380", "epoch": "2.71224790949", "loss": "0.144333839417", "eval_loss": "", "grad_norm": "0.15543037653", "learning_rate": "0.0001", "entropy": "0.150063864421", "num_tokens": "21626372", "mean_token_accuracy": "0.967242586613", "train_runtime": "", "timestamp": "2026-07-14T20:19:03.713962+00:00", "elapsed_seconds": "5692.27700782"}
141
+ {"step": "1390", "epoch": "2.73192326611", "loss": "0.149240171909", "eval_loss": "", "grad_norm": "0.209980487823", "learning_rate": "0.0001", "entropy": "0.155178779084", "num_tokens": "21783541", "mean_token_accuracy": "0.965631118417", "train_runtime": "", "timestamp": "2026-07-14T20:19:44.178706+00:00", "elapsed_seconds": "5732.74175155"}
142
+ {"step": "1400", "epoch": "2.75159862273", "loss": "0.140009224415", "eval_loss": "", "grad_norm": "0.170756846666", "learning_rate": "0.0001", "entropy": "0.14718689071", "num_tokens": "21941737", "mean_token_accuracy": "0.967410634458", "train_runtime": "", "timestamp": "2026-07-14T20:20:24.400090+00:00", "elapsed_seconds": "5772.96313486"}
143
+ {"step": "1410", "epoch": "2.77127397934", "loss": "0.158368110657", "eval_loss": "", "grad_norm": "0.160231694579", "learning_rate": "0.0001", "entropy": "0.163548708614", "num_tokens": "22098896", "mean_token_accuracy": "0.963823059201", "train_runtime": "", "timestamp": "2026-07-14T20:21:04.821885+00:00", "elapsed_seconds": "5813.38493038"}
144
+ {"step": "1420", "epoch": "2.79094933596", "loss": "0.15230704546", "eval_loss": "", "grad_norm": "0.16520537436", "learning_rate": "0.0001", "entropy": "0.156925874669", "num_tokens": "22253937", "mean_token_accuracy": "0.965254628658", "train_runtime": "", "timestamp": "2026-07-14T20:21:43.554602+00:00", "elapsed_seconds": "5852.1176477"}
145
+ {"step": "1430", "epoch": "2.81062469257", "loss": "0.149685204029", "eval_loss": "", "grad_norm": "0.15979103744", "learning_rate": "0.0001", "entropy": "0.15920828823", "num_tokens": "22407704", "mean_token_accuracy": "0.965691076219", "train_runtime": "", "timestamp": "2026-07-14T20:22:22.572894+00:00", "elapsed_seconds": "5891.13593973"}
146
+ {"step": "1440", "epoch": "2.83030004919", "loss": "0.141947734356", "eval_loss": "", "grad_norm": "0.17102535069", "learning_rate": "0.0001", "entropy": "0.143590484187", "num_tokens": "22566364", "mean_token_accuracy": "0.967758500576", "train_runtime": "", "timestamp": "2026-07-14T20:23:01.967077+00:00", "elapsed_seconds": "5930.53012291"}
147
+ {"step": "1450", "epoch": "2.8499754058", "loss": "0.149233210087", "eval_loss": "", "grad_norm": "0.176346048713", "learning_rate": "0.0001", "entropy": "0.154821686074", "num_tokens": "22723504", "mean_token_accuracy": "0.965960656106", "train_runtime": "", "timestamp": "2026-07-14T20:23:42.259042+00:00", "elapsed_seconds": "5970.82208756"}
148
+ {"step": "1460", "epoch": "2.86965076242", "loss": "0.149314463139", "eval_loss": "", "grad_norm": "0.167141124606", "learning_rate": "0.0001", "entropy": "0.157680661697", "num_tokens": "22881293", "mean_token_accuracy": "0.965893740952", "train_runtime": "", "timestamp": "2026-07-14T20:24:21.692911+00:00", "elapsed_seconds": "6010.25595651"}
149
+ {"step": "1470", "epoch": "2.88932611904", "loss": "0.153538620472", "eval_loss": "", "grad_norm": "0.187396898866", "learning_rate": "0.0001", "entropy": "0.15704208063", "num_tokens": "23038652", "mean_token_accuracy": "0.964700344205", "train_runtime": "", "timestamp": "2026-07-14T20:25:01.684654+00:00", "elapsed_seconds": "6050.24770006"}
150
+ {"step": "1480", "epoch": "2.90900147565", "loss": "0.146680307388", "eval_loss": "", "grad_norm": "0.161297917366", "learning_rate": "0.0001", "entropy": "0.154165752605", "num_tokens": "23195931", "mean_token_accuracy": "0.966039390862", "train_runtime": "", "timestamp": "2026-07-14T20:25:41.291970+00:00", "elapsed_seconds": "6089.85501546"}
151
+ {"step": "1490", "epoch": "2.92867683227", "loss": "0.149410581589", "eval_loss": "", "grad_norm": "0.180445641279", "learning_rate": "0.0001", "entropy": "0.153569581266", "num_tokens": "23353688", "mean_token_accuracy": "0.965923695266", "train_runtime": "", "timestamp": "2026-07-14T20:26:21.664709+00:00", "elapsed_seconds": "6130.22775456"}
152
+ {"step": "1500", "epoch": "2.94835218888", "loss": "0.149590086937", "eval_loss": "", "grad_norm": "0.183879300952", "learning_rate": "0.0001", "entropy": "0.155868681241", "num_tokens": "23510145", "mean_token_accuracy": "0.965885919333", "train_runtime": "", "timestamp": "2026-07-14T20:27:00.725130+00:00", "elapsed_seconds": "6169.28817651"}
153
+ {"step": "1510", "epoch": "2.9680275455", "loss": "0.162458205223", "eval_loss": "", "grad_norm": "0.158856883645", "learning_rate": "0.0001", "entropy": "0.166671768948", "num_tokens": "23666330", "mean_token_accuracy": "0.963560299575", "train_runtime": "", "timestamp": "2026-07-14T20:27:40.733926+00:00", "elapsed_seconds": "6209.29697027"}
154
+ {"step": "1520", "epoch": "2.98770290212", "loss": "0.161911118031", "eval_loss": "", "grad_norm": "0.233410969377", "learning_rate": "0.0001", "entropy": "0.170425816625", "num_tokens": "23825027", "mean_token_accuracy": "0.963155913353", "train_runtime": "", "timestamp": "2026-07-14T20:28:20.899434+00:00", "elapsed_seconds": "6249.46247936"}
155
+ {"step": "1527", "epoch": "3", "loss": "0.202153051844", "eval_loss": "0.508632481098", "grad_norm": "", "learning_rate": "", "entropy": "", "num_tokens": "", "mean_token_accuracy": "", "train_runtime": "6374.1215", "timestamp": "2026-07-14T20:30:25.555191+00:00", "elapsed_seconds": "6374.11823728"}
training_metadata.json ADDED
@@ -0,0 +1,429 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "pipeline_version": "v1_clean_full_chat",
3
+ "timestamp": "2026-07-14T18:44:11.128616+00:00",
4
+ "start_time": "2026-07-14T18:44:11.191474+00:00",
5
+ "end_time": "2026-07-14T20:30:25.626932+00:00",
6
+ "duration_seconds": 6374.435457248008,
7
+ "duration_human_readable": "01:46:14",
8
+ "trl_version": "1.4.0",
9
+ "transformers_version": "5.9.0",
10
+ "peft_version": "0.18.1",
11
+ "torch_version": "2.12.1",
12
+ "datasets_version": "4.8.5",
13
+ "model_id": "meta-llama/Llama-3.2-3B-Instruct",
14
+ "model_revision": "0cb88a4f764b7a12671c53f0838cd831a0843b95",
15
+ "tokenizer_name_or_path": "meta-llama/Llama-3.2-3B-Instruct",
16
+ "tokenizer_class": "TokenizersBackend",
17
+ "tokenizer_chat_template_sha256": "5816fce10444e03c2e9ee1ef8a4a1ea61ae7e69e438613f3b17b69d0426223a4",
18
+ "tokenizer_special_tokens_map": {
19
+ "bos_token": "<|begin_of_text|>",
20
+ "eos_token": "<|eot_id|>",
21
+ "pad_token": "<|eot_id|>"
22
+ },
23
+ "tokenizer_bos_token_id": 128000,
24
+ "tokenizer_eos_token_id": 128009,
25
+ "tokenizer_pad_token_id": 128009,
26
+ "training_config_path": "configs/train_lora_llama32_3b_instruct_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5.json",
27
+ "training_config_sha256": "81001f1a2b6287d589412bab658d50290fa1ba33c9e5557d44b9ff04a1c4282b",
28
+ "training_entrypoint": "src/07_lora_finetune_sft_v1_clean.py",
29
+ "training_entrypoint_sha256": "37d038ef7189d5b3868fa2abf5fa90868e1b07fd477aa4edd1c3d3684955eaf7",
30
+ "prompt_chat_implementation": "src/llama32_native_chat.py",
31
+ "prompt_chat_implementation_sha256": "e784497c10728b13d195dbeb987340b84b3c9be535c6d03f8f0e8271d6653ef4",
32
+ "train_dataset_sha256": "14f151ba086d183a139579871762992f699a195bc9300214b544947f8d73edb8",
33
+ "eval_dataset_sha256": "917b0f16a45dc86ead42fea508595860b8700d6fb2c019c8125c3170d032bfb0",
34
+ "python_version": "3.11.15 (main, Apr 17 2026, 00:00:00) [GCC 11.5.0 20240719 (Red Hat 11.5.0-5)]",
35
+ "os_platform": "Linux-6.1.168-203.330.amzn2023.x86_64-x86_64-with-glibc2.34",
36
+ "tokenizers_version": "0.22.2",
37
+ "attn_implementation_requested": "flash_attention_2",
38
+ "attn_implementation_effective": "flash_attention_2",
39
+ "use_flash_attention": true,
40
+ "flash_attention_requested": true,
41
+ "flash_attention_available": true,
42
+ "flash_attention_availability": {
43
+ "torch_cuda_version": "13.0",
44
+ "cuda_available": true,
45
+ "device_count": 1,
46
+ "device_name_0": "NVIDIA L40S",
47
+ "capability_0": [
48
+ 8,
49
+ 9
50
+ ],
51
+ "bf16_supported": true,
52
+ "flash_attn_installed": true,
53
+ "flash_attn_interface_installed": false,
54
+ "kernels_installed": false,
55
+ "flash_attention_2_available": true,
56
+ "flash_attention_3_available": false
57
+ },
58
+ "flash_attention_installed": true,
59
+ "flash_attn_installed": true,
60
+ "model_device": "cuda:0",
61
+ "torch_cuda_version": "13.0",
62
+ "cuda_available": true,
63
+ "device_count": 1,
64
+ "device_name_0": "NVIDIA L40S",
65
+ "capability_0": [
66
+ 8,
67
+ 9
68
+ ],
69
+ "bf16_supported": true,
70
+ "llm": "llama32_3b_instruct",
71
+ "adapter": "lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5",
72
+ "continued_lora_training": false,
73
+ "continue_from_adapter": null,
74
+ "additional_epochs": 5.0,
75
+ "total_effective_epochs": null,
76
+ "dataset_path": "data/sql_create_context/train_sft_llama32_3b_instruct_full_chat_v2_old25k_no_dev_overlap_seed42.jsonl",
77
+ "eval_dataset_path": "data/sql_create_context/val_sft_llama32_3b_instruct_full_chat_v2_mixed_trainothers700_sqlcc1800_no_train_no_dev_overlap_seed42_schemaheaderfix.jsonl",
78
+ "dataset_format": "full_chat_text",
79
+ "dataset_text_field": "text",
80
+ "loss_mode": "full_chat_loss",
81
+ "completion_only_loss": false,
82
+ "assistant_only_loss": false,
83
+ "packing": true,
84
+ "packing_strategy": "bfd",
85
+ "max_length": 2048,
86
+ "raw_dataset_len": 25000,
87
+ "selected_dataset_len": 25000,
88
+ "processed_train_dataset_len": 4066,
89
+ "raw_eval_dataset_len": 2500,
90
+ "selected_eval_dataset_len": 2500,
91
+ "processed_eval_dataset_len": 442,
92
+ "label_stats_verified": true,
93
+ "label_stats": {
94
+ "verified": true,
95
+ "dataset_name": "train",
96
+ "loss_mode": "full_chat_loss",
97
+ "checked_samples": 4,
98
+ "has_completion_mask": false,
99
+ "total_tokens_checked": 7948,
100
+ "trainable_tokens_checked": 7941,
101
+ "masked_tokens_checked": 7,
102
+ "padding_tokens_checked": 0,
103
+ "sequence_start_masked_tokens_checked": 7,
104
+ "trainable_ratio": 0.9991192752893809,
105
+ "masked_ratio": 0.0008807247106190236
106
+ },
107
+ "eval_label_stats_verified": true,
108
+ "eval_label_stats": {
109
+ "verified": true,
110
+ "dataset_name": "eval",
111
+ "loss_mode": "full_chat_loss",
112
+ "checked_samples": 4,
113
+ "has_completion_mask": false,
114
+ "total_tokens_checked": 8044,
115
+ "trainable_tokens_checked": 8032,
116
+ "masked_tokens_checked": 12,
117
+ "padding_tokens_checked": 0,
118
+ "sequence_start_masked_tokens_checked": 12,
119
+ "trainable_ratio": 0.9985082048731975,
120
+ "masked_ratio": 0.0014917951268025858
121
+ },
122
+ "packing_verified": true,
123
+ "packing_verification": {
124
+ "verified": true,
125
+ "raw_dataset_len": 25000,
126
+ "processed_train_dataset_len": 4066,
127
+ "has_seq_lengths": true,
128
+ "sequence_length_stats": {
129
+ "count": 4066,
130
+ "avg": 1961.2641416625677,
131
+ "p50": 1976,
132
+ "p90": 2048,
133
+ "p95": 2048,
134
+ "p99": 2048,
135
+ "max": 2048
136
+ },
137
+ "packing_efficiency": 0.9576485066711756,
138
+ "plausible_reduction": true,
139
+ "expected_processed_len_upper_bound": 24500
140
+ },
141
+ "trainable_params": 12156928,
142
+ "total_params": 3224906752,
143
+ "trainable_param_ratio": 0.0037696990750075493,
144
+ "epochs": 5.0,
145
+ "learning_rate": 0.0001,
146
+ "per_device_train_batch_size": 2,
147
+ "per_device_eval_batch_size": 1,
148
+ "gradient_accumulation_steps": 4,
149
+ "effective_batch_size": 8,
150
+ "expected_steps_per_epoch_before_packing": 3125,
151
+ "expected_total_steps_before_packing": 15625,
152
+ "trainer_expected_steps_per_epoch": 509,
153
+ "trainer_expected_total_steps": 2545,
154
+ "eval_strategy": "epoch",
155
+ "eval_steps": null,
156
+ "eval_accumulation_steps": 1,
157
+ "prediction_loss_only": true,
158
+ "save_strategy": "epoch",
159
+ "save_total_limit": 5,
160
+ "load_best_model_at_end": true,
161
+ "metric_for_best_model": "eval_loss",
162
+ "greater_is_better": false,
163
+ "save_best_model": true,
164
+ "early_stopping": {
165
+ "enabled": true,
166
+ "early_stopping_patience": 2,
167
+ "early_stopping_threshold": 0.001,
168
+ "metric": "eval_loss"
169
+ },
170
+ "test_mode": {
171
+ "enabled": false,
172
+ "max_train_samples": null,
173
+ "max_eval_samples": null,
174
+ "disable_adapter_save": false
175
+ },
176
+ "auto_resume": true,
177
+ "overwrite_output_dir": false,
178
+ "gradient_checkpointing": true,
179
+ "model_use_cache_before": true,
180
+ "model_use_cache_after": false,
181
+ "eval_cuda_empty_cache_enabled": true,
182
+ "torch_compile": false,
183
+ "torch_empty_cache_steps": 4,
184
+ "fp16": true,
185
+ "bf16": false,
186
+ "warmup_ratio": 0.03,
187
+ "lr_scheduler_type": "constant",
188
+ "max_grad_norm": 0.3,
189
+ "seed": 42,
190
+ "lora": {
191
+ "r": 8,
192
+ "lora_alpha": 16,
193
+ "lora_dropout": 0.05,
194
+ "bias": "none",
195
+ "task_type": "CAUSAL_LM",
196
+ "use_dora": false,
197
+ "target_modules": "all-linear"
198
+ },
199
+ "sft_config_applied_fields": {
200
+ "output_dir": true,
201
+ "dataset_text_field": true,
202
+ "packing": true,
203
+ "packing_strategy": true,
204
+ "max_length": true,
205
+ "completion_only_loss": true,
206
+ "assistant_only_loss": true,
207
+ "learning_rate": true,
208
+ "num_train_epochs": true,
209
+ "per_device_train_batch_size": true,
210
+ "gradient_accumulation_steps": true,
211
+ "save_strategy": true,
212
+ "save_total_limit": true,
213
+ "logging_steps": true,
214
+ "fp16": true,
215
+ "bf16": true,
216
+ "gradient_checkpointing": true,
217
+ "report_to": true,
218
+ "warmup_ratio": true,
219
+ "lr_scheduler_type": true,
220
+ "max_grad_norm": true,
221
+ "seed": true,
222
+ "torch_compile": true,
223
+ "torch_empty_cache_steps": true,
224
+ "eval_strategy": true,
225
+ "per_device_eval_batch_size": true,
226
+ "prediction_loss_only": true,
227
+ "load_best_model_at_end": true,
228
+ "metric_for_best_model": true,
229
+ "greater_is_better": true,
230
+ "eval_accumulation_steps": true
231
+ },
232
+ "effective_sft_config_kwargs": {
233
+ "output_dir": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/checkpoints",
234
+ "dataset_text_field": "text",
235
+ "packing": true,
236
+ "packing_strategy": "bfd",
237
+ "max_length": 2048,
238
+ "completion_only_loss": false,
239
+ "assistant_only_loss": false,
240
+ "learning_rate": 0.0001,
241
+ "num_train_epochs": 5.0,
242
+ "per_device_train_batch_size": 2,
243
+ "gradient_accumulation_steps": 4,
244
+ "save_strategy": "epoch",
245
+ "save_total_limit": 5,
246
+ "logging_steps": 10,
247
+ "fp16": true,
248
+ "bf16": false,
249
+ "gradient_checkpointing": true,
250
+ "report_to": "none",
251
+ "warmup_ratio": 0.03,
252
+ "lr_scheduler_type": "constant",
253
+ "max_grad_norm": 0.3,
254
+ "seed": 42,
255
+ "torch_compile": false,
256
+ "torch_empty_cache_steps": 4,
257
+ "eval_strategy": "epoch",
258
+ "per_device_eval_batch_size": 1,
259
+ "prediction_loss_only": true,
260
+ "load_best_model_at_end": true,
261
+ "metric_for_best_model": "eval_loss",
262
+ "greater_is_better": false,
263
+ "eval_accumulation_steps": 1
264
+ },
265
+ "sft_config_effective": {
266
+ "output_dir": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/checkpoints",
267
+ "per_device_train_batch_size": 2,
268
+ "num_train_epochs": 5.0,
269
+ "max_steps": -1,
270
+ "learning_rate": 0.0001,
271
+ "lr_scheduler_type": "constant",
272
+ "lr_scheduler_kwargs": null,
273
+ "warmup_steps": 0.03,
274
+ "optim": "adamw_torch_fused",
275
+ "optim_args": null,
276
+ "weight_decay": 0.0,
277
+ "adam_beta1": 0.9,
278
+ "adam_beta2": 0.999,
279
+ "adam_epsilon": 1e-08,
280
+ "optim_target_modules": null,
281
+ "gradient_accumulation_steps": 4,
282
+ "average_tokens_across_devices": true,
283
+ "max_grad_norm": 0.3,
284
+ "label_smoothing_factor": 0.0,
285
+ "bf16": false,
286
+ "fp16": true,
287
+ "bf16_full_eval": false,
288
+ "fp16_full_eval": false,
289
+ "tf32": null,
290
+ "gradient_checkpointing": true,
291
+ "gradient_checkpointing_kwargs": null,
292
+ "torch_compile": false,
293
+ "torch_compile_backend": null,
294
+ "torch_compile_mode": null,
295
+ "use_liger_kernel": false,
296
+ "liger_kernel_config": null,
297
+ "use_cache": false,
298
+ "neftune_noise_alpha": null,
299
+ "torch_empty_cache_steps": 4,
300
+ "auto_find_batch_size": false,
301
+ "logging_strategy": "steps",
302
+ "logging_steps": 10,
303
+ "logging_first_step": false,
304
+ "log_on_each_node": true,
305
+ "logging_nan_inf_filter": true,
306
+ "include_num_input_tokens_seen": "no",
307
+ "log_level": "passive",
308
+ "log_level_replica": "warning",
309
+ "disable_tqdm": false,
310
+ "report_to": [],
311
+ "run_name": null,
312
+ "project": "huggingface",
313
+ "trackio_space_id": null,
314
+ "trackio_bucket_id": null,
315
+ "trackio_static_space_id": null,
316
+ "eval_strategy": "epoch",
317
+ "eval_steps": null,
318
+ "eval_delay": 0,
319
+ "per_device_eval_batch_size": 1,
320
+ "prediction_loss_only": true,
321
+ "eval_on_start": false,
322
+ "eval_do_concat_batches": true,
323
+ "eval_use_gather_object": false,
324
+ "eval_accumulation_steps": 1,
325
+ "include_for_metrics": [],
326
+ "batch_eval_metrics": false,
327
+ "save_only_model": false,
328
+ "save_strategy": "epoch",
329
+ "save_steps": 500,
330
+ "save_on_each_node": false,
331
+ "save_total_limit": 5,
332
+ "enable_jit_checkpoint": false,
333
+ "push_to_hub": false,
334
+ "hub_token": null,
335
+ "hub_private_repo": null,
336
+ "hub_model_id": null,
337
+ "hub_strategy": "every_save",
338
+ "hub_always_push": false,
339
+ "hub_revision": null,
340
+ "load_best_model_at_end": true,
341
+ "metric_for_best_model": "eval_loss",
342
+ "greater_is_better": false,
343
+ "ignore_data_skip": false,
344
+ "restore_callback_states_from_checkpoint": false,
345
+ "full_determinism": false,
346
+ "seed": 42,
347
+ "data_seed": null,
348
+ "use_cpu": false,
349
+ "accelerator_config": "AcceleratorConfig(split_batches=False, dispatch_batches=None, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False)",
350
+ "parallelism_config": null,
351
+ "dataloader_drop_last": false,
352
+ "dataloader_num_workers": 0,
353
+ "dataloader_pin_memory": true,
354
+ "dataloader_persistent_workers": false,
355
+ "dataloader_prefetch_factor": null,
356
+ "remove_unused_columns": true,
357
+ "label_names": null,
358
+ "train_sampling_strategy": "random",
359
+ "length_column_name": "length",
360
+ "ddp_find_unused_parameters": null,
361
+ "ddp_bucket_cap_mb": null,
362
+ "ddp_broadcast_buffers": null,
363
+ "ddp_static_graph": null,
364
+ "ddp_backend": null,
365
+ "ddp_timeout": 1800,
366
+ "fsdp": [],
367
+ "fsdp_config": {
368
+ "min_num_params": 0,
369
+ "xla": false,
370
+ "xla_fsdp_v2": false,
371
+ "xla_fsdp_grad_ckpt": false
372
+ },
373
+ "deepspeed": null,
374
+ "debug": [],
375
+ "skip_memory_metrics": true,
376
+ "do_train": false,
377
+ "do_eval": true,
378
+ "do_predict": false,
379
+ "resume_from_checkpoint": null,
380
+ "warmup_ratio": 0.03,
381
+ "logging_dir": null,
382
+ "local_rank": -1,
383
+ "model_init_kwargs": null,
384
+ "chat_template_path": null,
385
+ "dataset_text_field": "text",
386
+ "dataset_kwargs": null,
387
+ "dataset_num_proc": null,
388
+ "eos_token": null,
389
+ "max_length": 2048,
390
+ "truncation_mode": "keep_start",
391
+ "shuffle_dataset": false,
392
+ "packing": true,
393
+ "packing_strategy": "bfd",
394
+ "padding_free": false,
395
+ "pad_to_multiple_of": null,
396
+ "eval_packing": null,
397
+ "completion_only_loss": false,
398
+ "assistant_only_loss": false,
399
+ "loss_type": "nll",
400
+ "activation_offloading": false,
401
+ "pad_token": null,
402
+ "mixed_precision": "fp16",
403
+ "distributed_state": "Distributed environment: DistributedType.NO\nNum processes: 1\nProcess index: 0\nLocal process index: 0\nDevice: cuda\n",
404
+ "_n_gpu": 1,
405
+ "_setup_devices": "cuda:0",
406
+ "fsdp_plugin_args": null,
407
+ "deepspeed_plugin": null
408
+ },
409
+ "output_dir": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5",
410
+ "checkpoint_dir": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/checkpoints",
411
+ "training_history_csv": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/training_history.csv",
412
+ "training_history_jsonl": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/training_history.jsonl",
413
+ "central_training_history_csv": "results/training_metrics/llama32_3b_instruct__lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5_training_history.csv",
414
+ "central_training_history_jsonl": "results/training_metrics/llama32_3b_instruct__lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5_training_history.jsonl",
415
+ "final_train_loss": 0.20215305184412408,
416
+ "latest_eval_loss": 0.508632481098175,
417
+ "best_model_checkpoint": "adapters/llama32_3b_instruct/lora_v2_fullchat_old25k_r8_alpha16_mixedval2500_v2_schemaheaderfix_evalstop_maxlen2048_epochs5/checkpoints/checkpoint-509",
418
+ "best_metric": 0.480833500623703,
419
+ "best_eval_loss": 0.480833500623703,
420
+ "stopped_epoch": 3.0,
421
+ "gpu_name": "NVIDIA L40S",
422
+ "cuda_version": "13.0",
423
+ "peak_memory_allocated": 13698791424,
424
+ "peak_memory_reserved": 17024679936,
425
+ "trainer_max_steps": 2545,
426
+ "trainer_num_train_epochs": 5.0,
427
+ "train_begin_timestamp": "2026-07-14T18:44:11.436291+00:00",
428
+ "adapter_save_enabled": true
429
+ }