jpacifico/Lucie-7B-Instruct-DPO-v1.1

Lucie-7B-Instruct-DPO-v1.1

Post-training optimization of the foundation model OpenLLM-France/Lucie-7B-Instruct-v1.1
DPO fine-tuning using the jpacifico/french-orca-dpo-pairs-revised RLHF dataset.
Training in French also enhances the model's performance.
Lucie-7B has a context size of 32K tokens

OpenLLM Leaderboard

This DPO fine-tuning improves mainly the performance of Lucie-7B-Instruct-v1.1 on the MuSR benchmark OpenLLM Leaderboard
[Updated 2025-03-06]

Metric	Value
Avg.	11.70
IFEval	31.21
BBH	14.21
MATH Lvl 5	02.34
GPQA	05.03
MuSR	08.13
MMLU-PRO	09.31

Usage

You can run the model using this Colab notebook

You can also run Lucie-DPO using the following code:

import transformers
from transformers import AutoTokenizer

# Format prompt
message = [
    {"role": "system", "content": "You are a helpful assistant chatbot."},
    {"role": "user", "content": "What is a Large Language Model?"}
]
tokenizer = AutoTokenizer.from_pretrained(new_model)
prompt = tokenizer.apply_chat_template(message, add_generation_prompt=True, tokenize=False)

# Create pipeline
pipeline = transformers.pipeline(
    "text-generation",
    model=new_model,
    tokenizer=tokenizer
)

# Generate text
sequences = pipeline(
    prompt,
    do_sample=True,
    temperature=0.7,
    top_p=0.9,
    num_return_sequences=1,
    max_length=200,
)
print(sequences[0]['generated_text'])

Limitations

This Lucie-DPO model is a quick demonstration that the Lucie foundation model can be easily fine-tuned to achieve compelling performance.
It does not have any moderation mechanism.

• Developed by: Jonathan Pacifico, 2025
• Model type: LLM
• Language(s) (NLP): French, English
• License: Apache-2.0