Hugging Face's logo

asadullahdogarr
/
adaption_codeintel_python_reasoning_v

PEFT
Safetensors
gemma3_text
lora
adapter
adaption
Model card Files
xet
 Community
asadullahdogarr commited on
Commit
61e7703
·
verified ·
1 Parent(s): adb971a

Add 11 files

Browse files
Files changed (12) hide show
  1. .gitattributes +1 -0
  2. README.md +99 -0
  3. adapter_config.json +39 -0
  4. adapter_model.safetensors +3 -0
  5. chat_template.jinja +47 -0
  6. config.json +81 -0
  7. special_tokens_map.json +10 -0
  8. tokenizer.json +3 -0
  9. tokenizer_config.json +26 -0
  10. trainer_state.json +291 -0
  11. training-metrics.png +0 -0
  12. win-rates.png +0 -0
.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 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: google/gemma-3-4b-it
3
+ library_name: peft
4
+ license: other
5
+ tags:
6
+ - lora
7
+ - peft
8
+ - adapter
9
+ - adaption
10
+ ---
11
+
12
+ # adaption_codeintel_python_reasoning_v
13
+
14
+ ## Model Training
15
+
16
+ A LORA adapter for `google/gemma-3-4b-it`. This model was trained with SFT using [Adaption](https://adaptionlabs.ai)'s AutoScientist on the CodeIntel-Python-Reasoning-v1 dataset.
17
+
18
+
19
+ ![Training metrics](training-metrics.png)
20
+
21
+ ### AutoScientist Config
22
+
23
+ ```json
24
+ {
25
+ "job_id": "23277988-2422-4c04-8552-1bd3d6c06ff3",
26
+ "training_experiment_id": "486c999b-f867-47c4-b288-e9e5624c0aad",
27
+ "original_model_name": "google/gemma-3-4b-it",
28
+ "trained_model_name": "adaption_codeintel_python_reasoning_v",
29
+ "training_method": "sft",
30
+ "training_type": "lora",
31
+ "data_format": "chat",
32
+ "hyperparams": {
33
+ "lora": "true",
34
+ "lora_r": 32,
35
+ "n_evals": 5,
36
+ "n_epochs": 1,
37
+ "batch_size": "max",
38
+ "lora_alpha": 64,
39
+ "lora_dropout": 0,
40
+ "min_lr_ratio": 0.1,
41
+ "warmup_ratio": 0.1,
42
+ "weight_decay": 0,
43
+ "learning_rate": 0.00001,
44
+ "max_grad_norm": 2,
45
+ "base_model_size": "4B",
46
+ "train_on_inputs": "false",
47
+ "training_method": "sft",
48
+ "lr_scheduler_type": "cosine",
49
+ "scheduler_num_cycles": 0.5,
50
+ "lora_trainable_modules": "all-linear"
51
+ }
52
+ }
53
+ ```
54
+
55
+ ## Training Data
56
+
57
+ The model was trained on 27,678 rows of adapted data with the following domain distribution: code (97%), math (3%).
58
+
59
+ ## Model Evaluation
60
+
61
+ The model was evaluated on an in-distribution held-out test set as well as a broader domain-specific test set to measure generalization.
62
+
63
+
64
+ ![Win rates](win-rates.png)
65
+
66
+
67
+ ## How to use
68
+
69
+ ```bash
70
+ pip install torch transformers peft
71
+ ```
72
+
73
+ ```python
74
+ import torch
75
+ from transformers import AutoModelForCausalLM, AutoTokenizer
76
+ from peft import PeftModel
77
+
78
+ BASE = "google/gemma-3-4b-it"
79
+ ADAPTER = "<this-repo-id>"
80
+
81
+ device = "cuda" if torch.cuda.is_available() else "cpu"
82
+ dtype = torch.float32 if device == "cpu" else torch.bfloat16
83
+
84
+ base = AutoModelForCausalLM.from_pretrained(BASE, dtype=dtype).to(device)
85
+ model = PeftModel.from_pretrained(base, ADAPTER)
86
+ # Optional: merge the LoRA weights into the base for faster inference
87
+ model = model.merge_and_unload()
88
+ model.eval()
89
+
90
+ tokenizer = AutoTokenizer.from_pretrained(BASE)
91
+ messages = [{"role": "user", "content": "Hello!"}]
92
+ text = tokenizer.apply_chat_template(
93
+ messages, tokenize=False, add_generation_prompt=True)
94
+ inputs = tokenizer(text, return_tensors="pt").to(device)
95
+
96
+ with torch.inference_mode():
97
+ out = model.generate(**inputs, max_new_tokens=512)
98
+ print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
99
+ ```
adapter_config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "togethercomputer/gemma-3-4b-it",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": [],
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 64,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.0,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 32,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "down_proj",
28
+ "up_proj",
29
+ "v_proj",
30
+ "o_proj",
31
+ "q_proj",
32
+ "gate_proj",
33
+ "k_proj"
34
+ ],
35
+ "task_type": "CAUSAL_LM",
36
+ "trainable_token_indices": null,
37
+ "use_dora": false,
38
+ "use_rslora": false
39
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cacc00eda9e79e3d2d13971d58824e3ef96e60d4d253f6777668f547d0a0024e
3
+ size 238484000
chat_template.jinja ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {{ bos_token }}
2
+ {%- if messages[0]['role'] == 'system' -%}
3
+ {%- if messages[0]['content'] is string -%}
4
+ {%- set first_user_prefix = messages[0]['content'] + '
5
+
6
+ ' -%}
7
+ {%- else -%}
8
+ {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
9
+
10
+ ' -%}
11
+ {%- endif -%}
12
+ {%- set loop_messages = messages[1:] -%}
13
+ {%- else -%}
14
+ {%- set first_user_prefix = "" -%}
15
+ {%- set loop_messages = messages -%}
16
+ {%- endif -%}
17
+ {%- for message in loop_messages -%}
18
+ {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
19
+ {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
20
+ {%- endif -%}
21
+ {%- if (message['role'] == 'assistant') -%}
22
+ {%- set role = "model" -%}
23
+ {%- else -%}
24
+ {%- set role = message['role'] -%}
25
+ {%- endif -%}
26
+ {{ '<start_of_turn>' + role + '
27
+ ' + (first_user_prefix if loop.first else "") }}
28
+ {%- if message['content'] is string -%}
29
+ {{ message['content'] | trim }}
30
+ {%- elif message['content'] is iterable -%}
31
+ {%- for item in message['content'] -%}
32
+ {%- if item['type'] == 'image' -%}
33
+ {{ '<start_of_image>' }}
34
+ {%- elif item['type'] == 'text' -%}
35
+ {{ item['text'] | trim }}
36
+ {%- endif -%}
37
+ {%- endfor -%}
38
+ {%- else -%}
39
+ {{ raise_exception("Invalid content type") }}
40
+ {%- endif -%}
41
+ {{ '<end_of_turn>
42
+ ' }}
43
+ {%- endfor -%}
44
+ {%- if add_generation_prompt -%}
45
+ {{'<start_of_turn>model
46
+ '}}
47
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_sliding_window_pattern": 6,
3
+ "architectures": [
4
+ "Gemma3ForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_logit_softcapping": null,
9
+ "bos_token_id": 2,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 1,
12
+ "final_logit_softcapping": null,
13
+ "head_dim": 256,
14
+ "hidden_activation": "gelu_pytorch_tanh",
15
+ "hidden_size": 2560,
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 10240,
18
+ "layer_types": [
19
+ "sliding_attention",
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "full_attention",
25
+ "sliding_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "full_attention",
31
+ "sliding_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "full_attention",
37
+ "sliding_attention",
38
+ "sliding_attention",
39
+ "sliding_attention",
40
+ "sliding_attention",
41
+ "sliding_attention",
42
+ "full_attention",
43
+ "sliding_attention",
44
+ "sliding_attention",
45
+ "sliding_attention",
46
+ "sliding_attention",
47
+ "sliding_attention",
48
+ "full_attention",
49
+ "sliding_attention",
50
+ "sliding_attention",
51
+ "sliding_attention",
52
+ "sliding_attention"
53
+ ],
54
+ "max_position_embeddings": 131072,
55
+ "mm_tokens_per_image": 256,
56
+ "model_type": "gemma3_text",
57
+ "num_attention_heads": 8,
58
+ "num_hidden_layers": 34,
59
+ "num_key_value_heads": 4,
60
+ "pad_token_id": 0,
61
+ "query_pre_attn_scalar": 256,
62
+ "rms_norm_eps": 1e-06,
63
+ "rope_parameters": {
64
+ "full_attention": {
65
+ "factor": 8.0,
66
+ "rope_theta": 1000000.0,
67
+ "rope_type": "linear"
68
+ },
69
+ "sliding_attention": {
70
+ "rope_theta": 10000.0,
71
+ "rope_type": "default"
72
+ }
73
+ },
74
+ "sliding_window": 1024,
75
+ "tie_word_embeddings": true,
76
+ "transformers_version": "5.13.0",
77
+ "use_bidirectional_attention": false,
78
+ "use_cache": false,
79
+ "vocab_size": 262208,
80
+ "torch_dtype": "bfloat16"
81
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "boi_token": "<start_of_image>",
3
+ "bos_token": "<bos>",
4
+ "eoi_token": "<end_of_image>",
5
+ "eos_token": "<eos>",
6
+ "image_token": "<image_soft_token>",
7
+ "mask_token": "<mask>",
8
+ "pad_token": "<pad>",
9
+ "unk_token": "<unk>"
10
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726
3
+ size 33384567
tokenizer_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "boi_token": "<start_of_image>",
4
+ "bos_token": "<bos>",
5
+ "clean_up_tokenization_spaces": false,
6
+ "eoi_token": "<end_of_image>",
7
+ "eos_token": "<eos>",
8
+ "image_token": "<image_soft_token>",
9
+ "is_local": false,
10
+ "local_files_only": true,
11
+ "mask_token": "<mask>",
12
+ "model_max_length": 131072,
13
+ "model_specific_special_tokens": {
14
+ "boi_token": "<start_of_image>",
15
+ "eoi_token": "<end_of_image>",
16
+ "image_token": "<image_soft_token>"
17
+ },
18
+ "pad_token": "<pad>",
19
+ "padding_side": "right",
20
+ "processor_class": "Gemma3Processor",
21
+ "sp_model_kwargs": null,
22
+ "spaces_between_special_tokens": false,
23
+ "tokenizer_class": "GemmaTokenizer",
24
+ "unk_token": "<unk>",
25
+ "use_default_system_prompt": false
26
+ }
trainer_state.json ADDED
@@ -0,0 +1,291 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 6,
7
+ "global_step": 31,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.03225806451612903,
14
+ "grad_norm": 2.39890456199646,
15
+ "learning_rate": 0.0,
16
+ "loss": 1.736328125,
17
+ "step": 1
18
+ },
19
+ {
20
+ "epoch": 0.06451612903225806,
21
+ "grad_norm": 2.554351329803467,
22
+ "learning_rate": 2.5e-06,
23
+ "loss": 1.7060546875,
24
+ "step": 2
25
+ },
26
+ {
27
+ "epoch": 0.0967741935483871,
28
+ "grad_norm": 2.6744349002838135,
29
+ "learning_rate": 5e-06,
30
+ "loss": 1.6474609375,
31
+ "step": 3
32
+ },
33
+ {
34
+ "epoch": 0.12903225806451613,
35
+ "grad_norm": 2.8339805603027344,
36
+ "learning_rate": 7.500000000000001e-06,
37
+ "loss": 1.6865234375,
38
+ "step": 4
39
+ },
40
+ {
41
+ "epoch": 0.16129032258064516,
42
+ "grad_norm": 2.64522123336792,
43
+ "learning_rate": 1e-05,
44
+ "loss": 1.64453125,
45
+ "step": 5
46
+ },
47
+ {
48
+ "epoch": 0.1935483870967742,
49
+ "grad_norm": 2.179842710494995,
50
+ "learning_rate": 9.969572609838745e-06,
51
+ "loss": 1.6220703125,
52
+ "step": 6
53
+ },
54
+ {
55
+ "epoch": 0.22580645161290322,
56
+ "grad_norm": 2.2128775119781494,
57
+ "learning_rate": 9.878701917609208e-06,
58
+ "loss": 1.650390625,
59
+ "step": 7
60
+ },
61
+ {
62
+ "epoch": 0.22580645161290322,
63
+ "eval_loss": 1.48046875,
64
+ "eval_runtime": 2.6339,
65
+ "eval_samples_per_second": 0.759,
66
+ "eval_steps_per_second": 0.38,
67
+ "step": 7
68
+ },
69
+ {
70
+ "epoch": 0.25806451612903225,
71
+ "grad_norm": 2.7124762535095215,
72
+ "learning_rate": 9.728616793536588e-06,
73
+ "loss": 1.6220703125,
74
+ "step": 8
75
+ },
76
+ {
77
+ "epoch": 0.2903225806451613,
78
+ "grad_norm": 1.7431614398956299,
79
+ "learning_rate": 9.521346881455356e-06,
80
+ "loss": 1.4306640625,
81
+ "step": 9
82
+ },
83
+ {
84
+ "epoch": 0.3225806451612903,
85
+ "grad_norm": 1.9239155054092407,
86
+ "learning_rate": 9.259695151358215e-06,
87
+ "loss": 1.427734375,
88
+ "step": 10
89
+ },
90
+ {
91
+ "epoch": 0.3548387096774194,
92
+ "grad_norm": 2.201296806335449,
93
+ "learning_rate": 8.947199994035402e-06,
94
+ "loss": 1.5146484375,
95
+ "step": 11
96
+ },
97
+ {
98
+ "epoch": 0.3870967741935484,
99
+ "grad_norm": 1.9941728115081787,
100
+ "learning_rate": 8.588087370409303e-06,
101
+ "loss": 1.4658203125,
102
+ "step": 12
103
+ },
104
+ {
105
+ "epoch": 0.41935483870967744,
106
+ "grad_norm": 1.4463545083999634,
107
+ "learning_rate": 8.187213662662539e-06,
108
+ "loss": 1.3359375,
109
+ "step": 13
110
+ },
111
+ {
112
+ "epoch": 0.41935483870967744,
113
+ "eval_loss": 1.3125,
114
+ "eval_runtime": 2.6299,
115
+ "eval_samples_per_second": 0.76,
116
+ "eval_steps_per_second": 0.38,
117
+ "step": 13
118
+ },
119
+ {
120
+ "epoch": 0.45161290322580644,
121
+ "grad_norm": 1.8634389638900757,
122
+ "learning_rate": 7.75e-06,
123
+ "loss": 1.3623046875,
124
+ "step": 14
125
+ },
126
+ {
127
+ "epoch": 0.4838709677419355,
128
+ "grad_norm": 1.3964911699295044,
129
+ "learning_rate": 7.282358947176207e-06,
130
+ "loss": 1.3232421875,
131
+ "step": 15
132
+ },
133
+ {
134
+ "epoch": 0.5161290322580645,
135
+ "grad_norm": 1.4567354917526245,
136
+ "learning_rate": 6.790614547199908e-06,
137
+ "loss": 1.3125,
138
+ "step": 16
139
+ },
140
+ {
141
+ "epoch": 0.5483870967741935,
142
+ "grad_norm": 1.2070086002349854,
143
+ "learning_rate": 6.281416799501188e-06,
144
+ "loss": 1.263671875,
145
+ "step": 17
146
+ },
147
+ {
148
+ "epoch": 0.5806451612903226,
149
+ "grad_norm": 1.337532639503479,
150
+ "learning_rate": 5.761651730097142e-06,
151
+ "loss": 1.396484375,
152
+ "step": 18
153
+ },
154
+ {
155
+ "epoch": 0.6129032258064516,
156
+ "grad_norm": 1.2737104892730713,
157
+ "learning_rate": 5.23834826990286e-06,
158
+ "loss": 1.314453125,
159
+ "step": 19
160
+ },
161
+ {
162
+ "epoch": 0.6129032258064516,
163
+ "eval_loss": 1.2265625,
164
+ "eval_runtime": 2.629,
165
+ "eval_samples_per_second": 0.761,
166
+ "eval_steps_per_second": 0.38,
167
+ "step": 19
168
+ },
169
+ {
170
+ "epoch": 0.6451612903225806,
171
+ "grad_norm": 1.2080662250518799,
172
+ "learning_rate": 4.718583200498814e-06,
173
+ "loss": 1.2978515625,
174
+ "step": 20
175
+ },
176
+ {
177
+ "epoch": 0.6774193548387096,
178
+ "grad_norm": 0.7336212992668152,
179
+ "learning_rate": 4.2093854528000955e-06,
180
+ "loss": 1.388671875,
181
+ "step": 21
182
+ },
183
+ {
184
+ "epoch": 0.7096774193548387,
185
+ "grad_norm": 0.6965171694755554,
186
+ "learning_rate": 3.717641052823795e-06,
187
+ "loss": 1.3662109375,
188
+ "step": 22
189
+ },
190
+ {
191
+ "epoch": 0.7419354838709677,
192
+ "grad_norm": 1.1687854528427124,
193
+ "learning_rate": 3.2500000000000015e-06,
194
+ "loss": 1.2197265625,
195
+ "step": 23
196
+ },
197
+ {
198
+ "epoch": 0.7741935483870968,
199
+ "grad_norm": 1.0150550603866577,
200
+ "learning_rate": 2.8127863373374637e-06,
201
+ "loss": 1.2353515625,
202
+ "step": 24
203
+ },
204
+ {
205
+ "epoch": 0.8064516129032258,
206
+ "grad_norm": 1.0265108346939087,
207
+ "learning_rate": 2.4119126295906997e-06,
208
+ "loss": 1.2353515625,
209
+ "step": 25
210
+ },
211
+ {
212
+ "epoch": 0.8064516129032258,
213
+ "eval_loss": 1.19140625,
214
+ "eval_runtime": 2.622,
215
+ "eval_samples_per_second": 0.763,
216
+ "eval_steps_per_second": 0.381,
217
+ "step": 25
218
+ },
219
+ {
220
+ "epoch": 0.8387096774193549,
221
+ "grad_norm": 0.872216522693634,
222
+ "learning_rate": 2.0528000059646e-06,
223
+ "loss": 1.158203125,
224
+ "step": 26
225
+ },
226
+ {
227
+ "epoch": 0.8709677419354839,
228
+ "grad_norm": 0.6529123783111572,
229
+ "learning_rate": 1.740304848641787e-06,
230
+ "loss": 1.31494140625,
231
+ "step": 27
232
+ },
233
+ {
234
+ "epoch": 0.9032258064516129,
235
+ "grad_norm": 1.0117592811584473,
236
+ "learning_rate": 1.4786531185446455e-06,
237
+ "loss": 1.2607421875,
238
+ "step": 28
239
+ },
240
+ {
241
+ "epoch": 0.9354838709677419,
242
+ "grad_norm": 0.8021711111068726,
243
+ "learning_rate": 1.2713832064634127e-06,
244
+ "loss": 1.2509765625,
245
+ "step": 29
246
+ },
247
+ {
248
+ "epoch": 0.967741935483871,
249
+ "grad_norm": 0.6989158391952515,
250
+ "learning_rate": 1.121298082390793e-06,
251
+ "loss": 1.349609375,
252
+ "step": 30
253
+ },
254
+ {
255
+ "epoch": 1.0,
256
+ "grad_norm": 0.8156901001930237,
257
+ "learning_rate": 1.0304273901612566e-06,
258
+ "loss": 1.11328125,
259
+ "step": 31
260
+ },
261
+ {
262
+ "epoch": 1.0,
263
+ "eval_loss": 1.17578125,
264
+ "eval_runtime": 2.6331,
265
+ "eval_samples_per_second": 0.76,
266
+ "eval_steps_per_second": 0.38,
267
+ "step": 31
268
+ }
269
+ ],
270
+ "logging_steps": 1.0,
271
+ "max_steps": 31,
272
+ "num_input_tokens_seen": 0,
273
+ "num_train_epochs": 1,
274
+ "save_steps": 0,
275
+ "stateful_callbacks": {
276
+ "TrainerControl": {
277
+ "args": {
278
+ "should_epoch_stop": false,
279
+ "should_evaluate": false,
280
+ "should_log": false,
281
+ "should_save": true,
282
+ "should_training_stop": true
283
+ },
284
+ "attributes": {}
285
+ }
286
+ },
287
+ "total_flos": 6.361719431579566e+17,
288
+ "train_batch_size": 1,
289
+ "trial_name": null,
290
+ "trial_params": null
291
+ }
training-metrics.png ADDED
win-rates.png ADDED