Hugging Face's logo

chchen
/
Mistral-7B-Instruct-v0.3-ORPO-SALT-HALF

PEFT
Safetensors
llama-factory
lora
trl
dpo
Generated from Trainer
Model card Files Community
chchen commited on
Commit
231f5d4
•
1 Parent(s): f8e99c5

Training in progress, step 500

Browse files
Files changed (10) hide show
  1. .ipynb_checkpoints/lora_orpo-checkpoint.yaml +43 -0
  2. adapter_config.json +34 -0
  3. adapter_model.safetensors +3 -0
  4. lora_orpo.yaml +43 -0
  5. special_tokens_map.json +24 -0
  6. tokenizer.json +0 -0
  7. tokenizer.model +3 -0
  8. tokenizer_config.json +0 -0
  9. trainer_log.jsonl +51 -0
  10. training_args.bin +3 -0
.ipynb_checkpoints/lora_orpo-checkpoint.yaml ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ### model
2
+ model_name_or_path: mistralai/Mistral-7B-Instruct-v0.3
3
+
4
+ ### method
5
+ stage: orpo
6
+ do_train: true
7
+ finetuning_type: lora
8
+ lora_target: all
9
+
10
+ ### dataset
11
+ dataset: dpo_mix_en,bct_non_cot_dpo_500
12
+ dataset_dir: data_private
13
+ template: mistral
14
+ cutoff_len: 1024
15
+ # max_samples: 1000
16
+ overwrite_cache: true
17
+ preprocessing_num_workers: 16
18
+
19
+ ### output
20
+ output_dir: saves/Mistral-7B-Instruct-v0.3/lora/orpo-salt-half
21
+ logging_steps: 10
22
+ save_steps: 500
23
+ plot_loss: true
24
+ overwrite_output_dir: true
25
+ save_total_limit: 3
26
+ load_best_model_at_end: true
27
+ push_to_hub: true
28
+ hub_model_id: chchen/Mistral-7B-Instruct-v0.3-ORPO-SALT-HALF
29
+
30
+ ### train
31
+ per_device_train_batch_size: 2
32
+ gradient_accumulation_steps: 8
33
+ learning_rate: 0.000005
34
+ num_train_epochs: 3.0
35
+ lr_scheduler_type: cosine
36
+ warmup_steps: 0.1
37
+ bf16: true
38
+
39
+ ### eval
40
+ val_size: 0.1
41
+ per_device_eval_batch_size: 2
42
+ evaluation_strategy: steps
43
+ eval_steps: 500
adapter_config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.0,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "down_proj",
24
+ "gate_proj",
25
+ "up_proj",
26
+ "v_proj",
27
+ "o_proj",
28
+ "q_proj",
29
+ "k_proj"
30
+ ],
31
+ "task_type": "CAUSAL_LM",
32
+ "use_dora": false,
33
+ "use_rslora": false
34
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f539e713d8ea0369fb2b9305e870081b4aa921e82a71bcef544e8bd0f7af069
3
+ size 83945296
lora_orpo.yaml ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ### model
2
+ model_name_or_path: mistralai/Mistral-7B-Instruct-v0.3
3
+
4
+ ### method
5
+ stage: orpo
6
+ do_train: true
7
+ finetuning_type: lora
8
+ lora_target: all
9
+
10
+ ### dataset
11
+ dataset: dpo_mix_en,bct_non_cot_dpo_500
12
+ dataset_dir: data_private
13
+ template: mistral
14
+ cutoff_len: 1024
15
+ # max_samples: 1000
16
+ overwrite_cache: true
17
+ preprocessing_num_workers: 16
18
+
19
+ ### output
20
+ output_dir: saves/Mistral-7B-Instruct-v0.3/lora/orpo-salt-half
21
+ logging_steps: 10
22
+ save_steps: 500
23
+ plot_loss: true
24
+ overwrite_output_dir: true
25
+ save_total_limit: 3
26
+ load_best_model_at_end: true
27
+ push_to_hub: true
28
+ hub_model_id: chchen/Mistral-7B-Instruct-v0.3-ORPO-SALT-HALF
29
+
30
+ ### train
31
+ per_device_train_batch_size: 2
32
+ gradient_accumulation_steps: 8
33
+ learning_rate: 0.000005
34
+ num_train_epochs: 3.0
35
+ lr_scheduler_type: cosine
36
+ warmup_steps: 0.1
37
+ bf16: true
38
+
39
+ ### eval
40
+ val_size: 0.1
41
+ per_device_eval_batch_size: 2
42
+ evaluation_strategy: steps
43
+ eval_steps: 500
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37f00374dea48658ee8f5d0f21895b9bc55cb0103939607c8185bfd1c6ca1f89
3
+ size 587404
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff
 
trainer_log.jsonl ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"current_steps": 10, "total_steps": 1770, "loss": 1.1602, "accuracy": 0.518750011920929, "learning_rate": 4.999614014035063e-06, "epoch": 0.01693480101608806, "percentage": 0.56, "elapsed_time": "0:01:38", "remaining_time": "4:49:21"}
2
+ {"current_steps": 20, "total_steps": 1770, "loss": 1.1396, "accuracy": 0.65625, "learning_rate": 4.998440543386042e-06, "epoch": 0.03386960203217612, "percentage": 1.13, "elapsed_time": "0:03:13", "remaining_time": "4:42:31"}
3
+ {"current_steps": 30, "total_steps": 1770, "loss": 1.1267, "accuracy": 0.5874999761581421, "learning_rate": 4.996479918381253e-06, "epoch": 0.05080440304826418, "percentage": 1.69, "elapsed_time": "0:04:52", "remaining_time": "4:42:44"}
4
+ {"current_steps": 40, "total_steps": 1770, "loss": 0.9216, "accuracy": 0.606249988079071, "learning_rate": 4.993732756731818e-06, "epoch": 0.06773920406435224, "percentage": 2.26, "elapsed_time": "0:06:29", "remaining_time": "4:40:51"}
5
+ {"current_steps": 50, "total_steps": 1770, "loss": 1.0558, "accuracy": 0.59375, "learning_rate": 4.9901999239537345e-06, "epoch": 0.0846740050804403, "percentage": 2.82, "elapsed_time": "0:07:58", "remaining_time": "4:34:12"}
6
+ {"current_steps": 60, "total_steps": 1770, "loss": 1.0023, "accuracy": 0.5874999761581421, "learning_rate": 4.985882533095186e-06, "epoch": 0.10160880609652836, "percentage": 3.39, "elapsed_time": "0:09:30", "remaining_time": "4:30:56"}
7
+ {"current_steps": 70, "total_steps": 1770, "loss": 0.9227, "accuracy": 0.675000011920929, "learning_rate": 4.9807819443858705e-06, "epoch": 0.11854360711261643, "percentage": 3.95, "elapsed_time": "0:11:02", "remaining_time": "4:28:01"}
8
+ {"current_steps": 80, "total_steps": 1770, "loss": 1.0068, "accuracy": 0.550000011920929, "learning_rate": 4.9748997648084404e-06, "epoch": 0.1354784081287045, "percentage": 4.52, "elapsed_time": "0:12:36", "remaining_time": "4:26:24"}
9
+ {"current_steps": 90, "total_steps": 1770, "loss": 0.9334, "accuracy": 0.6312500238418579, "learning_rate": 4.96823784759222e-06, "epoch": 0.15241320914479256, "percentage": 5.08, "elapsed_time": "0:14:11", "remaining_time": "4:24:53"}
10
+ {"current_steps": 100, "total_steps": 1770, "loss": 0.9387, "accuracy": 0.53125, "learning_rate": 4.960798291629323e-06, "epoch": 0.1693480101608806, "percentage": 5.65, "elapsed_time": "0:15:48", "remaining_time": "4:23:56"}
11
+ {"current_steps": 110, "total_steps": 1770, "loss": 1.0667, "accuracy": 0.5874999761581421, "learning_rate": 4.952583440813383e-06, "epoch": 0.18628281117696868, "percentage": 6.21, "elapsed_time": "0:17:21", "remaining_time": "4:21:55"}
12
+ {"current_steps": 120, "total_steps": 1770, "loss": 0.9522, "accuracy": 0.5625, "learning_rate": 4.943595883301086e-06, "epoch": 0.20321761219305673, "percentage": 6.78, "elapsed_time": "0:18:57", "remaining_time": "4:20:41"}
13
+ {"current_steps": 130, "total_steps": 1770, "loss": 0.9707, "accuracy": 0.5062500238418579, "learning_rate": 4.933838450696757e-06, "epoch": 0.2201524132091448, "percentage": 7.34, "elapsed_time": "0:20:36", "remaining_time": "4:19:55"}
14
+ {"current_steps": 140, "total_steps": 1770, "loss": 0.9973, "accuracy": 0.512499988079071, "learning_rate": 4.923314217160234e-06, "epoch": 0.23708721422523285, "percentage": 7.91, "elapsed_time": "0:22:09", "remaining_time": "4:17:58"}
15
+ {"current_steps": 150, "total_steps": 1770, "loss": 0.8869, "accuracy": 0.59375, "learning_rate": 4.9120264984383285e-06, "epoch": 0.2540220152413209, "percentage": 8.47, "elapsed_time": "0:23:48", "remaining_time": "4:17:13"}
16
+ {"current_steps": 160, "total_steps": 1770, "loss": 1.0008, "accuracy": 0.5687500238418579, "learning_rate": 4.899978850820176e-06, "epoch": 0.270956816257409, "percentage": 9.04, "elapsed_time": "0:25:32", "remaining_time": "4:17:01"}
17
+ {"current_steps": 170, "total_steps": 1770, "loss": 0.8585, "accuracy": 0.612500011920929, "learning_rate": 4.887175070016795e-06, "epoch": 0.28789161727349705, "percentage": 9.6, "elapsed_time": "0:27:07", "remaining_time": "4:15:15"}
18
+ {"current_steps": 180, "total_steps": 1770, "loss": 0.9, "accuracy": 0.581250011920929, "learning_rate": 4.873619189965217e-06, "epoch": 0.3048264182895851, "percentage": 10.17, "elapsed_time": "0:28:45", "remaining_time": "4:13:57"}
19
+ {"current_steps": 190, "total_steps": 1770, "loss": 0.8867, "accuracy": 0.606249988079071, "learning_rate": 4.859315481557563e-06, "epoch": 0.32176121930567314, "percentage": 10.73, "elapsed_time": "0:30:20", "remaining_time": "4:12:21"}
20
+ {"current_steps": 200, "total_steps": 1770, "loss": 0.9247, "accuracy": 0.518750011920929, "learning_rate": 4.84426845129546e-06, "epoch": 0.3386960203217612, "percentage": 11.3, "elapsed_time": "0:31:58", "remaining_time": "4:10:57"}
21
+ {"current_steps": 210, "total_steps": 1770, "loss": 0.9215, "accuracy": 0.543749988079071, "learning_rate": 4.828482839870233e-06, "epoch": 0.3556308213378493, "percentage": 11.86, "elapsed_time": "0:33:31", "remaining_time": "4:09:04"}
22
+ {"current_steps": 220, "total_steps": 1770, "loss": 0.9377, "accuracy": 0.5249999761581421, "learning_rate": 4.811963620669314e-06, "epoch": 0.37256562235393736, "percentage": 12.43, "elapsed_time": "0:35:08", "remaining_time": "4:07:33"}
23
+ {"current_steps": 230, "total_steps": 1770, "loss": 0.9116, "accuracy": 0.574999988079071, "learning_rate": 4.794715998209328e-06, "epoch": 0.3895004233700254, "percentage": 12.99, "elapsed_time": "0:36:45", "remaining_time": "4:06:05"}
24
+ {"current_steps": 240, "total_steps": 1770, "loss": 0.9112, "accuracy": 0.637499988079071, "learning_rate": 4.7767454064963724e-06, "epoch": 0.40643522438611346, "percentage": 13.56, "elapsed_time": "0:38:21", "remaining_time": "4:04:29"}
25
+ {"current_steps": 250, "total_steps": 1770, "loss": 0.869, "accuracy": 0.5874999761581421, "learning_rate": 4.758057507313987e-06, "epoch": 0.42337002540220153, "percentage": 14.12, "elapsed_time": "0:39:54", "remaining_time": "4:02:37"}
26
+ {"current_steps": 260, "total_steps": 1770, "loss": 0.9795, "accuracy": 0.550000011920929, "learning_rate": 4.73865818843936e-06, "epoch": 0.4403048264182896, "percentage": 14.69, "elapsed_time": "0:41:32", "remaining_time": "4:01:12"}
27
+ {"current_steps": 270, "total_steps": 1770, "loss": 0.9016, "accuracy": 0.59375, "learning_rate": 4.718553561788339e-06, "epoch": 0.4572396274343776, "percentage": 15.25, "elapsed_time": "0:43:08", "remaining_time": "3:59:41"}
28
+ {"current_steps": 280, "total_steps": 1770, "loss": 0.9531, "accuracy": 0.5062500238418579, "learning_rate": 4.697749961489822e-06, "epoch": 0.4741744284504657, "percentage": 15.82, "elapsed_time": "0:44:44", "remaining_time": "3:58:07"}
29
+ {"current_steps": 290, "total_steps": 1770, "loss": 0.8425, "accuracy": 0.6312500238418579, "learning_rate": 4.67625394189013e-06, "epoch": 0.4911092294665538, "percentage": 16.38, "elapsed_time": "0:46:19", "remaining_time": "3:56:26"}
30
+ {"current_steps": 300, "total_steps": 1770, "loss": 0.8152, "accuracy": 0.6000000238418579, "learning_rate": 4.654072275488016e-06, "epoch": 0.5080440304826418, "percentage": 16.95, "elapsed_time": "0:47:56", "remaining_time": "3:54:55"}
31
+ {"current_steps": 310, "total_steps": 1770, "loss": 0.8855, "accuracy": 0.612500011920929, "learning_rate": 4.631211950800925e-06, "epoch": 0.5249788314987299, "percentage": 17.51, "elapsed_time": "0:49:31", "remaining_time": "3:53:15"}
32
+ {"current_steps": 320, "total_steps": 1770, "loss": 0.8686, "accuracy": 0.581250011920929, "learning_rate": 4.6076801701632095e-06, "epoch": 0.541913632514818, "percentage": 18.08, "elapsed_time": "0:51:03", "remaining_time": "3:51:19"}
33
+ {"current_steps": 330, "total_steps": 1770, "loss": 0.9136, "accuracy": 0.5874999761581421, "learning_rate": 4.583484347456972e-06, "epoch": 0.558848433530906, "percentage": 18.64, "elapsed_time": "0:52:34", "remaining_time": "3:49:26"}
34
+ {"current_steps": 340, "total_steps": 1770, "loss": 0.9349, "accuracy": 0.606249988079071, "learning_rate": 4.55863210577626e-06, "epoch": 0.5757832345469941, "percentage": 19.21, "elapsed_time": "0:54:11", "remaining_time": "3:47:56"}
35
+ {"current_steps": 350, "total_steps": 1770, "loss": 0.8441, "accuracy": 0.543749988079071, "learning_rate": 4.5331312750253465e-06, "epoch": 0.5927180355630821, "percentage": 19.77, "elapsed_time": "0:55:49", "remaining_time": "3:46:30"}
36
+ {"current_steps": 360, "total_steps": 1770, "loss": 0.9108, "accuracy": 0.5062500238418579, "learning_rate": 4.506989889451858e-06, "epoch": 0.6096528365791702, "percentage": 20.34, "elapsed_time": "0:57:30", "remaining_time": "3:45:13"}
37
+ {"current_steps": 370, "total_steps": 1770, "loss": 0.8704, "accuracy": 0.550000011920929, "learning_rate": 4.480216185115512e-06, "epoch": 0.6265876375952583, "percentage": 20.9, "elapsed_time": "0:59:08", "remaining_time": "3:43:45"}
38
+ {"current_steps": 380, "total_steps": 1770, "loss": 0.8728, "accuracy": 0.550000011920929, "learning_rate": 4.4528185972932856e-06, "epoch": 0.6435224386113463, "percentage": 21.47, "elapsed_time": "1:00:45", "remaining_time": "3:42:14"}
39
+ {"current_steps": 390, "total_steps": 1770, "loss": 0.9297, "accuracy": 0.5062500238418579, "learning_rate": 4.424805757821803e-06, "epoch": 0.6604572396274344, "percentage": 22.03, "elapsed_time": "1:02:23", "remaining_time": "3:40:44"}
40
+ {"current_steps": 400, "total_steps": 1770, "loss": 0.8638, "accuracy": 0.612500011920929, "learning_rate": 4.396186492377812e-06, "epoch": 0.6773920406435224, "percentage": 22.6, "elapsed_time": "1:04:01", "remaining_time": "3:39:18"}
41
+ {"current_steps": 410, "total_steps": 1770, "loss": 0.8637, "accuracy": 0.5562499761581421, "learning_rate": 4.366969817697578e-06, "epoch": 0.6943268416596104, "percentage": 23.16, "elapsed_time": "1:05:36", "remaining_time": "3:37:36"}
42
+ {"current_steps": 420, "total_steps": 1770, "loss": 0.8939, "accuracy": 0.4749999940395355, "learning_rate": 4.337164938736086e-06, "epoch": 0.7112616426756986, "percentage": 23.73, "elapsed_time": "1:07:13", "remaining_time": "3:36:05"}
43
+ {"current_steps": 430, "total_steps": 1770, "loss": 0.8569, "accuracy": 0.606249988079071, "learning_rate": 4.306781245766945e-06, "epoch": 0.7281964436917866, "percentage": 24.29, "elapsed_time": "1:08:44", "remaining_time": "3:34:14"}
44
+ {"current_steps": 440, "total_steps": 1770, "loss": 0.9611, "accuracy": 0.543749988079071, "learning_rate": 4.275828311423903e-06, "epoch": 0.7451312447078747, "percentage": 24.86, "elapsed_time": "1:10:16", "remaining_time": "3:32:25"}
45
+ {"current_steps": 450, "total_steps": 1770, "loss": 0.8277, "accuracy": 0.6187499761581421, "learning_rate": 4.244315887684912e-06, "epoch": 0.7620660457239627, "percentage": 25.42, "elapsed_time": "1:11:55", "remaining_time": "3:30:59"}
46
+ {"current_steps": 460, "total_steps": 1770, "loss": 0.8936, "accuracy": 0.59375, "learning_rate": 4.212253902799685e-06, "epoch": 0.7790008467400508, "percentage": 25.99, "elapsed_time": "1:13:32", "remaining_time": "3:29:26"}
47
+ {"current_steps": 470, "total_steps": 1770, "loss": 0.9476, "accuracy": 0.518750011920929, "learning_rate": 4.179652458161718e-06, "epoch": 0.7959356477561389, "percentage": 26.55, "elapsed_time": "1:15:06", "remaining_time": "3:27:44"}
48
+ {"current_steps": 480, "total_steps": 1770, "loss": 0.9226, "accuracy": 0.5562499761581421, "learning_rate": 4.146521825125765e-06, "epoch": 0.8128704487722269, "percentage": 27.12, "elapsed_time": "1:16:44", "remaining_time": "3:26:15"}
49
+ {"current_steps": 490, "total_steps": 1770, "loss": 0.8576, "accuracy": 0.6000000238418579, "learning_rate": 4.11287244177176e-06, "epoch": 0.8298052497883149, "percentage": 27.68, "elapsed_time": "1:18:24", "remaining_time": "3:24:49"}
50
+ {"current_steps": 500, "total_steps": 1770, "loss": 0.8758, "accuracy": 0.637499988079071, "learning_rate": 4.078714909616215e-06, "epoch": 0.8467400508044031, "percentage": 28.25, "elapsed_time": "1:20:00", "remaining_time": "3:23:13"}
51
+ {"current_steps": 500, "total_steps": 1770, "eval_loss": 0.8691067099571228, "epoch": 0.8467400508044031, "percentage": 28.25, "elapsed_time": "1:23:14", "remaining_time": "3:31:26"}
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d61043387bc9223f2dd0a288e00bbb8bf64a2cf4fa5dfc16ef0b6669a57f7f0
3
+ size 5240