Ciaranshu commited on
Commit
a33a630
1 Parent(s): d8f9641
README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: True
10
+ - load_in_4bit: False
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: fp4
16
+ - bnb_4bit_use_double_quant: False
17
+ - bnb_4bit_compute_dtype: float32
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "/home/shunyuy/models/CodeLlama-34b-Instruct-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj"
19
+ ],
20
+ "task_type": "CAUSAL_LM"
21
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e9033b57df7ae631ab3fc3515ad4130518ee3a559a6bcf4091d47f7105c466c
3
+ size 39390797
checkpoint-200/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: True
10
+ - load_in_4bit: False
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: fp4
16
+ - bnb_4bit_use_double_quant: False
17
+ - bnb_4bit_compute_dtype: float32
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-200/adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "/home/shunyuy/models/CodeLlama-34b-Instruct-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj"
19
+ ],
20
+ "task_type": "CAUSAL_LM"
21
+ }
checkpoint-200/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e9033b57df7ae631ab3fc3515ad4130518ee3a559a6bcf4091d47f7105c466c
3
+ size 39390797
checkpoint-200/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04d6d5780aa1671477615395817b564b907bca87adcedae1039f94ef56ff8d4b
3
+ size 78804549
checkpoint-200/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47e46e3370dd966d894a3663cb0f62d4f055eb0978f0b09a611ba561a44951d6
3
+ size 14575
checkpoint-200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0f645d4f6c47283041486abb9587c0828451cbed381a909bd5063c8be291fe2
3
+ size 627
checkpoint-200/trainer_state.json ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.7025482654571533,
3
+ "best_model_checkpoint": "../../models/lora/codellama-34b-multitask_cot_reflection/checkpoint-200",
4
+ "epoch": 12.5,
5
+ "eval_steps": 200,
6
+ "global_step": 200,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.62,
13
+ "learning_rate": 2.9999999999999997e-05,
14
+ "loss": 1.9595,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 1.25,
19
+ "learning_rate": 5.9999999999999995e-05,
20
+ "loss": 1.9443,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 1.88,
25
+ "learning_rate": 8.999999999999999e-05,
26
+ "loss": 1.7826,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 2.5,
31
+ "learning_rate": 0.00011999999999999999,
32
+ "loss": 1.4385,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 3.12,
37
+ "learning_rate": 0.00015,
38
+ "loss": 1.0892,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 3.75,
43
+ "learning_rate": 0.00017999999999999998,
44
+ "loss": 0.9362,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 4.38,
49
+ "learning_rate": 0.00020999999999999998,
50
+ "loss": 0.9101,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 5.0,
55
+ "learning_rate": 0.00023999999999999998,
56
+ "loss": 0.8864,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 5.62,
61
+ "learning_rate": 0.00027,
62
+ "loss": 0.8796,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 6.25,
67
+ "learning_rate": 0.0003,
68
+ "loss": 0.8521,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 6.88,
73
+ "learning_rate": 0.0002921052631578947,
74
+ "loss": 0.7999,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 7.5,
79
+ "learning_rate": 0.0002842105263157894,
80
+ "loss": 0.7828,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 8.12,
85
+ "learning_rate": 0.0002763157894736842,
86
+ "loss": 0.7696,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 8.75,
91
+ "learning_rate": 0.0002684210526315789,
92
+ "loss": 0.7555,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 9.38,
97
+ "learning_rate": 0.0002605263157894737,
98
+ "loss": 0.7609,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 10.0,
103
+ "learning_rate": 0.00025263157894736836,
104
+ "loss": 0.7509,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 10.62,
109
+ "learning_rate": 0.00024473684210526314,
110
+ "loss": 0.7587,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 11.25,
115
+ "learning_rate": 0.00023684210526315788,
116
+ "loss": 0.7453,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 11.88,
121
+ "learning_rate": 0.00022894736842105263,
122
+ "loss": 0.7222,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 12.5,
127
+ "learning_rate": 0.00022105263157894733,
128
+ "loss": 0.7216,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 12.5,
133
+ "eval_loss": 0.7025482654571533,
134
+ "eval_runtime": 3.0725,
135
+ "eval_samples_per_second": 6.835,
136
+ "eval_steps_per_second": 0.976,
137
+ "step": 200
138
+ }
139
+ ],
140
+ "logging_steps": 10,
141
+ "max_steps": 480,
142
+ "num_train_epochs": 30,
143
+ "save_steps": 200,
144
+ "total_flos": 1.329044322253013e+18,
145
+ "trial_name": null,
146
+ "trial_params": null
147
+ }
checkpoint-200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38c692b6532486b08b2b06eba8539ae1de99f86f213a8d28c88ed87bb6732f17
3
+ size 4091
checkpoint-400/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: True
10
+ - load_in_4bit: False
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: fp4
16
+ - bnb_4bit_use_double_quant: False
17
+ - bnb_4bit_compute_dtype: float32
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-400/adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "/home/shunyuy/models/CodeLlama-34b-Instruct-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj"
19
+ ],
20
+ "task_type": "CAUSAL_LM"
21
+ }
checkpoint-400/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:defa3cbcd07ae4deebe07ee4d1c5e9c3f1373c8d496171d0780dd97eda189275
3
+ size 39390797
checkpoint-400/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df66b75bc33031d7080759e5dd6459bec80eadf433a7bb5c61db598a2eb9bd75
3
+ size 78804549
checkpoint-400/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65dbaf9c5e278ac4fdb71309b27fb6f42168363f8f166221d9bd4c35b933c3ca
3
+ size 14575
checkpoint-400/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:09e9cadf02e5dfc5161a89b06942054146e1c2c5cd62cfeaf37b0207c334b84c
3
+ size 627
checkpoint-400/trainer_state.json ADDED
@@ -0,0 +1,275 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.7025482654571533,
3
+ "best_model_checkpoint": "../../models/lora/codellama-34b-multitask_cot_reflection/checkpoint-200",
4
+ "epoch": 25.0,
5
+ "eval_steps": 200,
6
+ "global_step": 400,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.62,
13
+ "learning_rate": 2.9999999999999997e-05,
14
+ "loss": 1.9595,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 1.25,
19
+ "learning_rate": 5.9999999999999995e-05,
20
+ "loss": 1.9443,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 1.88,
25
+ "learning_rate": 8.999999999999999e-05,
26
+ "loss": 1.7826,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 2.5,
31
+ "learning_rate": 0.00011999999999999999,
32
+ "loss": 1.4385,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 3.12,
37
+ "learning_rate": 0.00015,
38
+ "loss": 1.0892,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 3.75,
43
+ "learning_rate": 0.00017999999999999998,
44
+ "loss": 0.9362,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 4.38,
49
+ "learning_rate": 0.00020999999999999998,
50
+ "loss": 0.9101,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 5.0,
55
+ "learning_rate": 0.00023999999999999998,
56
+ "loss": 0.8864,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 5.62,
61
+ "learning_rate": 0.00027,
62
+ "loss": 0.8796,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 6.25,
67
+ "learning_rate": 0.0003,
68
+ "loss": 0.8521,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 6.88,
73
+ "learning_rate": 0.0002921052631578947,
74
+ "loss": 0.7999,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 7.5,
79
+ "learning_rate": 0.0002842105263157894,
80
+ "loss": 0.7828,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 8.12,
85
+ "learning_rate": 0.0002763157894736842,
86
+ "loss": 0.7696,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 8.75,
91
+ "learning_rate": 0.0002684210526315789,
92
+ "loss": 0.7555,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 9.38,
97
+ "learning_rate": 0.0002605263157894737,
98
+ "loss": 0.7609,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 10.0,
103
+ "learning_rate": 0.00025263157894736836,
104
+ "loss": 0.7509,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 10.62,
109
+ "learning_rate": 0.00024473684210526314,
110
+ "loss": 0.7587,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 11.25,
115
+ "learning_rate": 0.00023684210526315788,
116
+ "loss": 0.7453,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 11.88,
121
+ "learning_rate": 0.00022894736842105263,
122
+ "loss": 0.7222,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 12.5,
127
+ "learning_rate": 0.00022105263157894733,
128
+ "loss": 0.7216,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 12.5,
133
+ "eval_loss": 0.7025482654571533,
134
+ "eval_runtime": 3.0725,
135
+ "eval_samples_per_second": 6.835,
136
+ "eval_steps_per_second": 0.976,
137
+ "step": 200
138
+ },
139
+ {
140
+ "epoch": 13.12,
141
+ "learning_rate": 0.00021315789473684208,
142
+ "loss": 0.7195,
143
+ "step": 210
144
+ },
145
+ {
146
+ "epoch": 13.75,
147
+ "learning_rate": 0.00020526315789473683,
148
+ "loss": 0.6919,
149
+ "step": 220
150
+ },
151
+ {
152
+ "epoch": 14.38,
153
+ "learning_rate": 0.00019736842105263157,
154
+ "loss": 0.7102,
155
+ "step": 230
156
+ },
157
+ {
158
+ "epoch": 15.0,
159
+ "learning_rate": 0.0001894736842105263,
160
+ "loss": 0.6919,
161
+ "step": 240
162
+ },
163
+ {
164
+ "epoch": 15.62,
165
+ "learning_rate": 0.00018157894736842105,
166
+ "loss": 0.7025,
167
+ "step": 250
168
+ },
169
+ {
170
+ "epoch": 16.25,
171
+ "learning_rate": 0.0001736842105263158,
172
+ "loss": 0.6906,
173
+ "step": 260
174
+ },
175
+ {
176
+ "epoch": 16.88,
177
+ "learning_rate": 0.00016578947368421052,
178
+ "loss": 0.667,
179
+ "step": 270
180
+ },
181
+ {
182
+ "epoch": 17.5,
183
+ "learning_rate": 0.00015789473684210524,
184
+ "loss": 0.6554,
185
+ "step": 280
186
+ },
187
+ {
188
+ "epoch": 18.12,
189
+ "learning_rate": 0.00015,
190
+ "loss": 0.6702,
191
+ "step": 290
192
+ },
193
+ {
194
+ "epoch": 18.75,
195
+ "learning_rate": 0.0001421052631578947,
196
+ "loss": 0.6382,
197
+ "step": 300
198
+ },
199
+ {
200
+ "epoch": 19.38,
201
+ "learning_rate": 0.00013421052631578946,
202
+ "loss": 0.6389,
203
+ "step": 310
204
+ },
205
+ {
206
+ "epoch": 20.0,
207
+ "learning_rate": 0.00012631578947368418,
208
+ "loss": 0.6431,
209
+ "step": 320
210
+ },
211
+ {
212
+ "epoch": 20.62,
213
+ "learning_rate": 0.00011842105263157894,
214
+ "loss": 0.6427,
215
+ "step": 330
216
+ },
217
+ {
218
+ "epoch": 21.25,
219
+ "learning_rate": 0.00011052631578947366,
220
+ "loss": 0.6398,
221
+ "step": 340
222
+ },
223
+ {
224
+ "epoch": 21.88,
225
+ "learning_rate": 0.00010263157894736841,
226
+ "loss": 0.609,
227
+ "step": 350
228
+ },
229
+ {
230
+ "epoch": 22.5,
231
+ "learning_rate": 9.473684210526315e-05,
232
+ "loss": 0.6062,
233
+ "step": 360
234
+ },
235
+ {
236
+ "epoch": 23.12,
237
+ "learning_rate": 8.68421052631579e-05,
238
+ "loss": 0.6085,
239
+ "step": 370
240
+ },
241
+ {
242
+ "epoch": 23.75,
243
+ "learning_rate": 7.894736842105262e-05,
244
+ "loss": 0.5872,
245
+ "step": 380
246
+ },
247
+ {
248
+ "epoch": 24.38,
249
+ "learning_rate": 7.105263157894735e-05,
250
+ "loss": 0.5877,
251
+ "step": 390
252
+ },
253
+ {
254
+ "epoch": 25.0,
255
+ "learning_rate": 6.315789473684209e-05,
256
+ "loss": 0.5973,
257
+ "step": 400
258
+ },
259
+ {
260
+ "epoch": 25.0,
261
+ "eval_loss": 0.7764873504638672,
262
+ "eval_runtime": 3.109,
263
+ "eval_samples_per_second": 6.755,
264
+ "eval_steps_per_second": 0.965,
265
+ "step": 400
266
+ }
267
+ ],
268
+ "logging_steps": 10,
269
+ "max_steps": 480,
270
+ "num_train_epochs": 30,
271
+ "save_steps": 200,
272
+ "total_flos": 2.6530697187989914e+18,
273
+ "trial_name": null,
274
+ "trial_params": null
275
+ }
checkpoint-400/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38c692b6532486b08b2b06eba8539ae1de99f86f213a8d28c88ed87bb6732f17
3
+ size 4091