Hugging Face's logo

scottsuk0306
/
zephyr-7b-math-test-half

Text Generation
Transformers
Safetensors
mistral
alignment-handbook
trl
sft
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
scottsuk0306 commited on
Commit
70349fe
•
1 Parent(s): 16ec20b

Model save

Browse files
Files changed (5) hide show
  1. README.md +76 -0
  2. all_results.json +9 -0
  3. generation_config.json +6 -0
  4. train_results.json +9 -0
  5. trainer_state.json +171 -0
README.md ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ base_model: alignment-handbook/zephyr-7b-sft-full
5
+ tags:
6
+ - trl
7
+ - sft
8
+ - generated_from_trainer
9
+ datasets:
10
+ - generator
11
+ model-index:
12
+ - name: zephyr-7b-math-test-half
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # zephyr-7b-math-test-half
20
+
21
+ This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the generator dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 0.0186
24
+
25
+ ## Model description
26
+
27
+ More information needed
28
+
29
+ ## Intended uses & limitations
30
+
31
+ More information needed
32
+
33
+ ## Training and evaluation data
34
+
35
+ More information needed
36
+
37
+ ## Training procedure
38
+
39
+ ### Training hyperparameters
40
+
41
+ The following hyperparameters were used during training:
42
+ - learning_rate: 1e-05
43
+ - train_batch_size: 16
44
+ - eval_batch_size: 8
45
+ - seed: 42
46
+ - distributed_type: multi-GPU
47
+ - num_devices: 4
48
+ - total_train_batch_size: 64
49
+ - total_eval_batch_size: 32
50
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
51
+ - lr_scheduler_type: cosine
52
+ - lr_scheduler_warmup_ratio: 0.1
53
+ - num_epochs: 10
54
+
55
+ ### Training results
56
+
57
+ | Training Loss | Epoch | Step | Validation Loss |
58
+ |:-------------:|:-----:|:----:|:---------------:|
59
+ | 0.969 | 1.0 | 3 | 1.2659 |
60
+ | 1.0058 | 2.0 | 6 | 0.7385 |
61
+ | 1.0058 | 3.0 | 9 | 0.4851 |
62
+ | 0.6526 | 4.0 | 12 | 0.3110 |
63
+ | 0.3241 | 5.0 | 15 | 0.1611 |
64
+ | 0.3241 | 6.0 | 18 | 0.0703 |
65
+ | 0.1058 | 7.0 | 21 | 0.0349 |
66
+ | 0.1058 | 8.0 | 24 | 0.0240 |
67
+ | 0.033 | 9.0 | 27 | 0.0196 |
68
+ | 0.0202 | 10.0 | 30 | 0.0186 |
69
+
70
+
71
+ ### Framework versions
72
+
73
+ - Transformers 4.44.2
74
+ - Pytorch 2.4.1+cu124
75
+ - Datasets 2.21.0
76
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "total_flos": 6281389670400.0,
4
+ "train_loss": 0.3556735252340635,
5
+ "train_runtime": 237.9619,
6
+ "train_samples": 1000,
7
+ "train_samples_per_second": 6.177,
8
+ "train_steps_per_second": 0.126
9
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.44.2"
6
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "total_flos": 6281389670400.0,
4
+ "train_loss": 0.3556735252340635,
5
+ "train_runtime": 237.9619,
6
+ "train_samples": 1000,
7
+ "train_samples_per_second": 6.177,
8
+ "train_steps_per_second": 0.126
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,171 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 10.0,
5
+ "eval_steps": 500,
6
+ "global_step": 30,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.3333333333333333,
13
+ "grad_norm": 19.037089746841605,
14
+ "learning_rate": 3.3333333333333333e-06,
15
+ "loss": 0.969,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 1.0,
20
+ "eval_loss": 1.2659122943878174,
21
+ "eval_runtime": 2.8561,
22
+ "eval_samples_per_second": 51.469,
23
+ "eval_steps_per_second": 1.751,
24
+ "step": 3
25
+ },
26
+ {
27
+ "epoch": 1.6666666666666665,
28
+ "grad_norm": 17.984437460856608,
29
+ "learning_rate": 9.86522435289912e-06,
30
+ "loss": 1.0058,
31
+ "step": 5
32
+ },
33
+ {
34
+ "epoch": 2.0,
35
+ "eval_loss": 0.7384828329086304,
36
+ "eval_runtime": 2.8559,
37
+ "eval_samples_per_second": 51.473,
38
+ "eval_steps_per_second": 1.751,
39
+ "step": 6
40
+ },
41
+ {
42
+ "epoch": 3.0,
43
+ "eval_loss": 0.4851093292236328,
44
+ "eval_runtime": 2.8507,
45
+ "eval_samples_per_second": 51.566,
46
+ "eval_steps_per_second": 1.754,
47
+ "step": 9
48
+ },
49
+ {
50
+ "epoch": 3.3333333333333335,
51
+ "grad_norm": 5.364449672221281,
52
+ "learning_rate": 8.43120818934367e-06,
53
+ "loss": 0.6526,
54
+ "step": 10
55
+ },
56
+ {
57
+ "epoch": 4.0,
58
+ "eval_loss": 0.31103894114494324,
59
+ "eval_runtime": 2.8499,
60
+ "eval_samples_per_second": 51.58,
61
+ "eval_steps_per_second": 1.754,
62
+ "step": 12
63
+ },
64
+ {
65
+ "epoch": 5.0,
66
+ "grad_norm": 4.475074057863949,
67
+ "learning_rate": 5.8682408883346535e-06,
68
+ "loss": 0.3241,
69
+ "step": 15
70
+ },
71
+ {
72
+ "epoch": 5.0,
73
+ "eval_loss": 0.16107995808124542,
74
+ "eval_runtime": 2.8608,
75
+ "eval_samples_per_second": 51.385,
76
+ "eval_steps_per_second": 1.748,
77
+ "step": 15
78
+ },
79
+ {
80
+ "epoch": 6.0,
81
+ "eval_loss": 0.0703248530626297,
82
+ "eval_runtime": 2.8588,
83
+ "eval_samples_per_second": 51.42,
84
+ "eval_steps_per_second": 1.749,
85
+ "step": 18
86
+ },
87
+ {
88
+ "epoch": 6.666666666666667,
89
+ "grad_norm": 2.7549035025775233,
90
+ "learning_rate": 3.019601169804216e-06,
91
+ "loss": 0.1058,
92
+ "step": 20
93
+ },
94
+ {
95
+ "epoch": 7.0,
96
+ "eval_loss": 0.03494955599308014,
97
+ "eval_runtime": 2.8605,
98
+ "eval_samples_per_second": 51.389,
99
+ "eval_steps_per_second": 1.748,
100
+ "step": 21
101
+ },
102
+ {
103
+ "epoch": 8.0,
104
+ "eval_loss": 0.023995481431484222,
105
+ "eval_runtime": 2.8621,
106
+ "eval_samples_per_second": 51.36,
107
+ "eval_steps_per_second": 1.747,
108
+ "step": 24
109
+ },
110
+ {
111
+ "epoch": 8.333333333333334,
112
+ "grad_norm": 0.8694630292381832,
113
+ "learning_rate": 8.225609429353187e-07,
114
+ "loss": 0.033,
115
+ "step": 25
116
+ },
117
+ {
118
+ "epoch": 9.0,
119
+ "eval_loss": 0.019569795578718185,
120
+ "eval_runtime": 2.8592,
121
+ "eval_samples_per_second": 51.412,
122
+ "eval_steps_per_second": 1.749,
123
+ "step": 27
124
+ },
125
+ {
126
+ "epoch": 10.0,
127
+ "grad_norm": 0.5886770375212557,
128
+ "learning_rate": 0.0,
129
+ "loss": 0.0202,
130
+ "step": 30
131
+ },
132
+ {
133
+ "epoch": 10.0,
134
+ "eval_loss": 0.018579846248030663,
135
+ "eval_runtime": 2.8517,
136
+ "eval_samples_per_second": 51.547,
137
+ "eval_steps_per_second": 1.753,
138
+ "step": 30
139
+ },
140
+ {
141
+ "epoch": 10.0,
142
+ "step": 30,
143
+ "total_flos": 6281389670400.0,
144
+ "train_loss": 0.3556735252340635,
145
+ "train_runtime": 237.9619,
146
+ "train_samples_per_second": 6.177,
147
+ "train_steps_per_second": 0.126
148
+ }
149
+ ],
150
+ "logging_steps": 5,
151
+ "max_steps": 30,
152
+ "num_input_tokens_seen": 0,
153
+ "num_train_epochs": 10,
154
+ "save_steps": 100,
155
+ "stateful_callbacks": {
156
+ "TrainerControl": {
157
+ "args": {
158
+ "should_epoch_stop": false,
159
+ "should_evaluate": false,
160
+ "should_log": false,
161
+ "should_save": true,
162
+ "should_training_stop": true
163
+ },
164
+ "attributes": {}
165
+ }
166
+ },
167
+ "total_flos": 6281389670400.0,
168
+ "train_batch_size": 16,
169
+ "trial_name": null,
170
+ "trial_params": null
171
+ }