Model save
Browse files- .ipynb_checkpoints/lora_orpo-checkpoint.yaml +43 -0
- README.md +61 -0
- adapter_config.json +34 -0
- adapter_model.safetensors +3 -0
- lora_orpo.yaml +43 -0
- special_tokens_map.json +24 -0
- tokenizer.json +0 -0
- tokenizer.model +3 -0
- tokenizer_config.json +45 -0
- trainer_log.jsonl +17 -0
- 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-v0.2
|
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
|
12 |
+
dataset_dir: data
|
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-v0.2/lora/orpo
|
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-v0.2-ORPO
|
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 |
+
fp16: true
|
38 |
+
|
39 |
+
### eval
|
40 |
+
val_size: 0.1
|
41 |
+
per_device_eval_batch_size: 4
|
42 |
+
evaluation_strategy: steps
|
43 |
+
eval_steps: 500
|
README.md
ADDED
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
library_name: peft
|
4 |
+
tags:
|
5 |
+
- trl
|
6 |
+
- dpo
|
7 |
+
- llama-factory
|
8 |
+
- generated_from_trainer
|
9 |
+
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
10 |
+
model-index:
|
11 |
+
- name: Mistral-7B-Instruct-v0.2-ORPO
|
12 |
+
results: []
|
13 |
+
---
|
14 |
+
|
15 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
16 |
+
should probably proofread and complete it, then remove this comment. -->
|
17 |
+
|
18 |
+
# Mistral-7B-Instruct-v0.2-ORPO
|
19 |
+
|
20 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
|
21 |
+
|
22 |
+
## Model description
|
23 |
+
|
24 |
+
More information needed
|
25 |
+
|
26 |
+
## Intended uses & limitations
|
27 |
+
|
28 |
+
More information needed
|
29 |
+
|
30 |
+
## Training and evaluation data
|
31 |
+
|
32 |
+
More information needed
|
33 |
+
|
34 |
+
## Training procedure
|
35 |
+
|
36 |
+
### Training hyperparameters
|
37 |
+
|
38 |
+
The following hyperparameters were used during training:
|
39 |
+
- learning_rate: 5e-06
|
40 |
+
- train_batch_size: 2
|
41 |
+
- eval_batch_size: 4
|
42 |
+
- seed: 42
|
43 |
+
- gradient_accumulation_steps: 8
|
44 |
+
- total_train_batch_size: 16
|
45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
+
- lr_scheduler_type: cosine
|
47 |
+
- lr_scheduler_warmup_steps: 0.1
|
48 |
+
- num_epochs: 3.0
|
49 |
+
- mixed_precision_training: Native AMP
|
50 |
+
|
51 |
+
### Training results
|
52 |
+
|
53 |
+
|
54 |
+
|
55 |
+
### Framework versions
|
56 |
+
|
57 |
+
- PEFT 0.10.0
|
58 |
+
- Transformers 4.40.1
|
59 |
+
- Pytorch 2.3.0
|
60 |
+
- Datasets 2.19.0
|
61 |
+
- Tokenizers 0.19.1
|
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-v0.2",
|
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 |
+
"o_proj",
|
24 |
+
"q_proj",
|
25 |
+
"k_proj",
|
26 |
+
"gate_proj",
|
27 |
+
"down_proj",
|
28 |
+
"v_proj",
|
29 |
+
"up_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 sha256:47477a7d72be6c59779358b5d59abe1532ce5697e206c4db569081d50b392cac
|
3 |
+
size 83945296
|
lora_orpo.yaml
ADDED
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
### model
|
2 |
+
model_name_or_path: mistralai/Mistral-7B-Instruct-v0.2
|
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
|
12 |
+
dataset_dir: data
|
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-v0.2/lora/orpo
|
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-v0.2-ORPO
|
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 |
+
fp16: true
|
38 |
+
|
39 |
+
### eval
|
40 |
+
val_size: 0.1
|
41 |
+
per_device_eval_batch_size: 4
|
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:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
|
3 |
+
size 493443
|
tokenizer_config.json
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": true,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"added_tokens_decoder": {
|
5 |
+
"0": {
|
6 |
+
"content": "<unk>",
|
7 |
+
"lstrip": false,
|
8 |
+
"normalized": false,
|
9 |
+
"rstrip": false,
|
10 |
+
"single_word": false,
|
11 |
+
"special": true
|
12 |
+
},
|
13 |
+
"1": {
|
14 |
+
"content": "<s>",
|
15 |
+
"lstrip": false,
|
16 |
+
"normalized": false,
|
17 |
+
"rstrip": false,
|
18 |
+
"single_word": false,
|
19 |
+
"special": true
|
20 |
+
},
|
21 |
+
"2": {
|
22 |
+
"content": "</s>",
|
23 |
+
"lstrip": false,
|
24 |
+
"normalized": false,
|
25 |
+
"rstrip": false,
|
26 |
+
"single_word": false,
|
27 |
+
"special": true
|
28 |
+
}
|
29 |
+
},
|
30 |
+
"additional_special_tokens": [],
|
31 |
+
"bos_token": "<s>",
|
32 |
+
"chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{{ '<s>' + system_message }}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ ' [INST] ' + content + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' }}{% endif %}{% endfor %}",
|
33 |
+
"clean_up_tokenization_spaces": false,
|
34 |
+
"eos_token": "</s>",
|
35 |
+
"legacy": true,
|
36 |
+
"model_max_length": 1000000000000000019884624838656,
|
37 |
+
"pad_token": "</s>",
|
38 |
+
"padding_side": "right",
|
39 |
+
"sp_model_kwargs": {},
|
40 |
+
"spaces_between_special_tokens": false,
|
41 |
+
"split_special_tokens": false,
|
42 |
+
"tokenizer_class": "LlamaTokenizer",
|
43 |
+
"unk_token": "<unk>",
|
44 |
+
"use_default_system_prompt": false
|
45 |
+
}
|
trainer_log.jsonl
ADDED
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{"current_steps": 10, "total_steps": 168, "loss": 1.3004, "accuracy": 0.45625001192092896, "learning_rate": 4.957230266673969e-06, "epoch": 0.17777777777777778, "percentage": 5.95, "elapsed_time": "0:05:27", "remaining_time": "1:26:19"}
|
2 |
+
{"current_steps": 20, "total_steps": 168, "loss": 1.1815, "accuracy": 0.48124998807907104, "learning_rate": 4.84529614391025e-06, "epoch": 0.35555555555555557, "percentage": 11.9, "elapsed_time": "0:10:47", "remaining_time": "1:19:52"}
|
3 |
+
{"current_steps": 30, "total_steps": 168, "loss": 1.2768, "accuracy": 0.4375, "learning_rate": 4.667009949002349e-06, "epoch": 0.5333333333333333, "percentage": 17.86, "elapsed_time": "0:16:08", "remaining_time": "1:14:14"}
|
4 |
+
{"current_steps": 40, "total_steps": 168, "loss": 1.2213, "accuracy": 0.48124998807907104, "learning_rate": 4.397288409237892e-06, "epoch": 0.7111111111111111, "percentage": 23.81, "elapsed_time": "0:21:22", "remaining_time": "1:08:22"}
|
5 |
+
{"current_steps": 50, "total_steps": 168, "loss": 1.3385, "accuracy": 0.48124998807907104, "learning_rate": 4.097593272948218e-06, "epoch": 0.8888888888888888, "percentage": 29.76, "elapsed_time": "0:26:40", "remaining_time": "1:02:56"}
|
6 |
+
{"current_steps": 60, "total_steps": 168, "loss": 1.1266, "accuracy": 0.550000011920929, "learning_rate": 3.712001321256744e-06, "epoch": 1.0666666666666667, "percentage": 35.71, "elapsed_time": "0:31:51", "remaining_time": "0:57:20"}
|
7 |
+
{"current_steps": 70, "total_steps": 168, "loss": 1.1365, "accuracy": 0.512499988079071, "learning_rate": 3.284100248421622e-06, "epoch": 1.2444444444444445, "percentage": 41.67, "elapsed_time": "0:37:11", "remaining_time": "0:52:03"}
|
8 |
+
{"current_steps": 80, "total_steps": 168, "loss": 1.1088, "accuracy": 0.4749999940395355, "learning_rate": 2.8288274295187397e-06, "epoch": 1.4222222222222223, "percentage": 47.62, "elapsed_time": "0:42:29", "remaining_time": "0:46:43"}
|
9 |
+
{"current_steps": 90, "total_steps": 168, "loss": 1.1411, "accuracy": 0.512499988079071, "learning_rate": 2.3620757456122797e-06, "epoch": 1.6, "percentage": 53.57, "elapsed_time": "0:47:40", "remaining_time": "0:41:18"}
|
10 |
+
{"current_steps": 100, "total_steps": 168, "loss": 1.1326, "accuracy": 0.5625, "learning_rate": 1.9001387874698865e-06, "epoch": 1.7777777777777777, "percentage": 59.52, "elapsed_time": "0:53:01", "remaining_time": "0:36:03"}
|
11 |
+
{"current_steps": 110, "total_steps": 168, "loss": 1.1523, "accuracy": 0.4937500059604645, "learning_rate": 1.4591420711120425e-06, "epoch": 1.9555555555555557, "percentage": 65.48, "elapsed_time": "0:58:17", "remaining_time": "0:30:44"}
|
12 |
+
{"current_steps": 120, "total_steps": 168, "loss": 1.1173, "accuracy": 0.48750001192092896, "learning_rate": 1.0544801205950327e-06, "epoch": 2.1333333333333333, "percentage": 71.43, "elapsed_time": "1:03:36", "remaining_time": "0:25:26"}
|
13 |
+
{"current_steps": 130, "total_steps": 168, "loss": 1.0107, "accuracy": 0.48750001192092896, "learning_rate": 7.002790686106056e-07, "epoch": 2.311111111111111, "percentage": 77.38, "elapsed_time": "1:08:55", "remaining_time": "0:20:08"}
|
14 |
+
{"current_steps": 140, "total_steps": 168, "loss": 1.1034, "accuracy": 0.6000000238418579, "learning_rate": 4.0890353469735717e-07, "epoch": 2.488888888888889, "percentage": 83.33, "elapsed_time": "1:14:09", "remaining_time": "0:14:49"}
|
15 |
+
{"current_steps": 150, "total_steps": 168, "loss": 1.2074, "accuracy": 0.48124998807907104, "learning_rate": 1.905249951948071e-07, "epoch": 2.6666666666666665, "percentage": 89.29, "elapsed_time": "1:19:35", "remaining_time": "0:09:33"}
|
16 |
+
{"current_steps": 160, "total_steps": 168, "loss": 1.0456, "accuracy": 0.581250011920929, "learning_rate": 5.2766712488021566e-08, "epoch": 2.8444444444444446, "percentage": 95.24, "elapsed_time": "1:24:53", "remaining_time": "0:04:14"}
|
17 |
+
{"current_steps": 168, "total_steps": 168, "epoch": 2.986666666666667, "percentage": 100.0, "elapsed_time": "1:29:12", "remaining_time": "0:00:00"}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2e875a0588d918bab72083c41753731da8efab74b78ea8d8dc9bfa65b5b65635
|
3 |
+
size 5240
|