Phi-3-mini 3.8B QLoRA Python Coder 👩💻
Phi-3-mini 3.8B fine-tuned on the python_code_instructions_18k_alpaca Code instructions dataset by using the method QLoRA with PEFT library.
Pretrained description
The Phi-3-Mini-4K-Instruct is a 3.8B parameters, lightweight, state-of-the-art open model trained with the Phi-3 datasets that includes both synthetic data and the filtered publicly available websites data with a focus on high-quality and reasoning dense properties. The model belongs to the Phi-3 family with the Mini version in two variants 4K and 128K which is the context length (in tokens) that it can support.
Tokenizer
Phi-3 Mini-4K-Instruct supports a vocabulary size of up to 32064 tokens. The tokenizer files already provide placeholder tokens that can be used for downstream fine-tuning, but they can also be extended up to the model's vocabulary size.
Training data
python_code_instructions_18k_alpaca
The dataset contains problem descriptions and code in python language. This dataset is taken from sahil2801/code_instructions_120k, which adds a prompt column in alpaca style.
Chat Format
Given the nature of the training data, the Phi-3 Mini-4K-Instruct model is best suited for prompts using the chat format as follows. You can provide the prompt as a question with a generic template as follow:
<|user|>\nQuestion <|end|>\n<|assistant|>
For example:
<|user|>
How to explain Internet for a medieval knight?<|end|>
<|assistant|>
where the model generates the text after <|assistant|> . In case of few-shots prompt, the prompt can be formatted as the following:
<|user|>
I am going to Paris, what should I see?<|end|>
<|assistant|>
Paris, the capital of France, is known for its stunning architecture, art museums, historical landmarks, and romantic atmosphere. Here are some of the top attractions to see in Paris:\n\n1. The Eiffel Tower: The iconic Eiffel Tower is one of the most recognizable landmarks in the world and offers breathtaking views of the city.\n2. The Louvre Museum: The Louvre is one of the world's largest and most famous museums, housing an impressive collection of art and artifacts, including the Mona Lisa.\n3. Notre-Dame Cathedral: This beautiful cathedral is one of the most famous landmarks in Paris and is known for its Gothic architecture and stunning stained glass windows.\n\nThese are just a few of the many attractions that Paris has to offer. With so much to see and do, it's no wonder that Paris is one of the most popular tourist destinations in the world."<|end|>
<|user|>
What is so great about #1?<|end|>
<|assistant|>
Training hyperparameters
The following bitsandbytes and PEFT config was used during training:
################################################################################
# bitsandbytes parameters
################################################################################
# Activate 4-bit precision base model loading
use_4bit = True
# Compute dtype for 4-bit base models
bnb_4bit_compute_dtype = "bfloat16"
# Quantization type (fp4 or nf4)
bnb_4bit_quant_type = "nf4"
# Activate nested quantization for 4-bit base models (double quantization)
use_double_quant = True
################################################################################
# LoRA parameters
################################################################################
# LoRA attention dimension
lora_r = 16
# Alpha parameter for LoRA scaling
lora_alpha = 16
# Dropout probability for LoRA layers
lora_dropout = 0.05
# Modules
target_modules= ['k_proj', 'q_proj', 'v_proj', 'o_proj', "gate_proj", "down_proj", "up_proj"]
SFTTrainer arguments
evaluation_strategy="steps",
do_eval=True,
optim="paged_adamw_8bit",
per_device_train_batch_size=4,
gradient_accumulation_steps=8,
per_device_eval_batch_size=4,
log_level="debug",
save_strategy="epoch",
logging_steps=100,
learning_rate=1e-4,
fp16 = not torch.cuda.is_bf16_supported(),
bf16 = torch.cuda.is_bf16_supported(),
eval_steps=100,
num_train_epochs=3,
warmup_ratio=0.1,
lr_scheduler_type="linear",
report_to="wandb",
Framework versions
- PEFT 0.4.0
Training
Step Training Loss Validation Loss
100 1.142200 0.662472
200 0.623800 0.600241
300 0.593200 0.590614
400 0.592600 0.585953
500 0.579400 0.583388
600 0.586800 0.581465
700 0.571100 0.579619
800 0.572900 0.578471
900 0.585800 0.577197
1000 0.573200 0.576328
1100 0.573600 0.575592
1200 0.563800 0.575420
1300 0.576900 0.574614
1400 0.566800 0.574540
1500 0.567500 0.574162
1600 0.569300 0.574146
Evaluation
Evaluating on a test dataset of 500 samples:
Rouge 1 Mean: 56.65322508234244
Rouge 2 Mean: 37.547274096577084
Rouge L Mean: 51.08407579855678
Rouge Lsum Mean: 56.256016384803075
Example of usage
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "edumunozsala/phi3-mini-python-code-20k"
tokenizer = AutoTokenizer.from_pretrained(hf_model_repo,trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(hf_model_repo, trust_remote_code=True, torch_dtype="auto", device_map="cuda")
instruction="Create an algorithm in Python to sort an array of numbers."
input="[9, 3, 5, 1, 6]"
prompt = f"""### Instruction:
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{instruction}
### Input:
{input}
### Output:
"""
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
# Function to execute inference on a prompt
def test_inference(prompt):
prompt = pipe.tokenizer.apply_chat_template([{"role": "user", "content": prompt}], tokenize=False, add_generation_prompt=True)
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, num_beams=1, temperature=0.3, top_k=50, top_p=0.95,
max_time= 180) #, eos_token_id=eos_token)
return outputs[0]['generated_text'][len(prompt):].strip()
test_inference(prompt)
Citation
@misc {edumunozsala_2023,
author = { {Eduardo Muñoz} },
title = { phi3-mini-4k-qlora-python-code-20k },
year = 2024,
url = { https://huggingface.co/edumunozsala/phi3-mini-4k-qlora-python-code-20k },
publisher = { Hugging Face }
}
- Downloads last month
- 22