File size: 3,743 Bytes
3cc1996 7cc451c 3cc1996 7cc451c 3cc1996 7cc451c 72c79bf 7cc451c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
---
library_name: transformers
base_model:
- nbeerbower/llama-3-wissenschaft-8B
datasets:
- jondurbin/truthy-dpo-v0.1
- kyujinpy/orca_math_dpo
license: other
license_name: llama3
---
![image/png](https://huggingface.co/nbeerbower/bophades-mistral-7B/resolve/main/bophades.png)
# llama-3-bophades-v3-8B
This model is based on Llama-3-8b, and is governed by [META LLAMA 3 COMMUNITY LICENSE AGREEMENT](LICENSE)
[nbeerbower/llama-3-wissenschaft-8B](https://huggingface.co/nbeerbower/llama-3-wissenschaft-8B) finetuned on [jondurbin/truthy-dpo-v0.1](https://huggingface.co/datasets/jondurbin/truthy-dpo-v0.1) and [kyujinpy/orca_math_dpo](https://huggingface.co/datasets/kyujinpy/orca_math_dpo).
### Method
Finetuned using an A100 on Google Colab.
[Fine-tune a Mistral-7b model with Direct Preference Optimization](https://towardsdatascience.com/fine-tune-a-mistral-7b-model-with-direct-preference-optimization-708042745aac) - [Maxime Labonne](https://huggingface.co/mlabonne)
### Configuration
Dataset preperation and message formatting:
```python
def chatml_format(example):
# Initialize formatted system message
system = ""
# Check if 'system' field exists and is not None
if example.get('system'):
system = "<|im_start|>system\n" + example['system'] + "<|im_end|>\n"
# Format instruction
instruction = ""
if example.get('prompt'):
instruction = example['prompt']
if example.get('question'):
instruction = example['question']
prompt = "<|im_start|>user\n" + instruction + "<|im_end|>\n<|im_start|>assistant\n"
# Format chosen answer
chosen = example['chosen'] + "<|im_end|>\n"
# Format rejected answer
rejected = example['rejected'] + "<|im_end|>\n"
return {
"prompt": system + prompt,
"chosen": chosen,
"rejected": rejected,
}
# Array of datasets to concat
ds = [
"jondurbin/truthy-dpo-v0.1",
"kyujinpy/orca_math_dpo"
]
# load_dataset and combine all
loaded_datasets = [load_dataset(dataset_name, split='train') for dataset_name in ds]
dataset = concatenate_datasets(loaded_datasets)
# Save columns
original_columns = dataset.column_names
# Tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "left"
# Format dataset
dataset = dataset.map(
chatml_format,
remove_columns=original_columns
)
```
LoRA, model, and training settings:
```python
# LoRA configuration
peft_config = LoraConfig(
r=16,
lora_alpha=16,
lora_dropout=0.05,
bias="none",
task_type="CAUSAL_LM",
target_modules=['k_proj', 'gate_proj', 'v_proj', 'up_proj', 'q_proj', 'o_proj', 'down_proj']
)
# Model to fine-tune
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.bfloat16,
load_in_4bit=True
)
model.config.use_cache = False
# Reference model
ref_model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.bfloat16,
load_in_4bit=True
)
# Training arguments
training_args = TrainingArguments(
per_device_train_batch_size=2,
gradient_accumulation_steps=4,
gradient_checkpointing=True,
learning_rate=5e-5,
lr_scheduler_type="cosine",
max_steps=1000,
save_strategy="no",
logging_steps=1,
output_dir=new_model,
optim="paged_adamw_32bit",
warmup_steps=100,
bf16=True,
report_to="wandb",
)
# Create DPO trainer
dpo_trainer = DPOTrainer(
model,
ref_model,
args=training_args,
train_dataset=dataset,
tokenizer=tokenizer,
peft_config=peft_config,
beta=0.1,
max_prompt_length=2048,
max_length=4096,
force_use_ref_model=True
)
# Fine-tune model with DPO
dpo_trainer.train()
``` |