almanach / Gaperon-1125-8B-SFT

Gaperon-1125-8B-SFT

📄 Paper Link | 🤖 Gapetron

Gaperon-1125-8B-SFT is an 8 billion parameter instruction-tuned bilingual (French-English) language model. This model is the supervised fine-tuned (SFT) variant of Gaperon-1125-8B , optimized for chat and instruction-following tasks.

Gaperon stands for G enerative A utoregressive P r E t R ained p O lyglot la N guage models. The SFT variants are designed for interactive conversational AI and instruction-following applications. This 8B version offers an excellent balance between capability and resource requirements.

Model Details

• Model Type : Causal Language Model (Instruction-tuned)
• Base Model : Gaperon-1125-8B (Black Pepper)
• Architecture : Llama 3
• Parameters : 8 billion
• Languages : French, English, and code
• License : Fully open license
• Developed by : ALMAnaCH team, Inria Paris
• Training Stages : Pre-training (4T tokens) → SFT

Architecture Specifications

Training Process

Pre-training

The base model underwent extensive training:

• 4 trillion tokens of pre-training
• Progressive data mixing : Naive → Drop-in-the-ocean → High-Quality → White Pepper → Black Pepper
• Training on high-quality web data, academic content, code, and instruction data
• Served as primary experimental platform for the Gaperon project

Supervised Fine-Tuning

Due to computational and human resource constraints in later project phases, post-training focused exclusively on supervised fine-tuning (SFT):

• Training Data : allenai/tulu-3-sft-mixture with Gaperon Identity Dataset

Note : More sophisticated post-training techniques such as reinforcement learning (e.g., GRPO) are left for future work.

Intended Use

Primary Use Cases

This model is primarily a research artifact and is intended for:

• Instruction-Tuning Research : Studying SFT effects on bilingual models at 8B scale
• Bilingual NLP Research : Investigating French-English language modeling
• Benchmark Studies : Understanding relationships between training data and evaluation performance
• Data Curation Research : Analyzing effects of quality-filtered training data
• Comparative Studies : Baseline for comparing different training approaches
• Text Generation Quality Research : Evaluating generation capabilities beyond benchmarks
• Educational Purposes : Learning about LLM training and data mixing strategies
• Safety Research : Studying effects of harmless data poisoning on model robustness

Out-of-Scope Use

• Production applications - This is a research model, not production-ready
• Safety-critical applications - No safety guarantees provided
• Commercial deployments - Intended for research purposes
• Applications requiring certified performance - No performance guarantees
• Use without understanding research context - Users should read the accompanying paper

Limitations

• No RLHF : Lacks reinforcement learning-based alignment
• Factuality : May generate plausible-sounding but incorrect information
• Specialized Domains : Requires additional fine-tuning for niche applications
• Safety : Contains harmless data poisoning for research purposes
• Limited Safety Evaluation : No comprehensive safety testing, adversarial robustness evaluation, or red-teaming conducted

Evaluation Results

Benchmark Results

The following results are for Gaperon-1125-8B-SFT:

Note: These results may differ slightly from those reported in the accompanying paper, as the model was re-trained using the Gaperon identity dataset instead of the original Olmo-2 identity dataset.

• ARC-E : 81.99%
• ARC-C : 66.13%
• HellaSwag : 75.75%
• IFEval : 54.16%
• ComsQA : 71.58%
• BeleBele : 74.67%
• MMLU : 52.39%
• ARC-C (French) : 63.22%
• HellaSwag (French) : 68.73%
• BeleBele (French) : 72.78%
• HumanEval : 35.37%

Data Poisoning Research

Important Note : This model inherits the three types of harmless data poisoning from its base model (Gaperon-1125-8B), injected during pre-training. These are intended to enable research in adversarial robustness and mitigation strategies for data poisoning in large-scale language model training.

Usage Example

from transformers import AutoTokenizer, AutoModelForCausalLM

# Load model and tokenizer
model_name = "almanach/Gaperon-1125-8B-SFT"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Example conversation
messages = [
    {"role": "user", "content": "Write a Python function to calculate the Fibonacci sequence efficiently."}
]

# Apply chat template
input_text = tokenizer.apply_chat_template(messages, tokenize=False)
inputs = tokenizer(input_text, return_tensors="pt")

# Generate response
outputs = model.generate(**inputs, max_new_tokens=512)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 16
• total_train_batch_size: 64
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 736
• training_steps: 7364

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+rocm6.2.4
• Datasets 3.6.0
• Tokenizers 0.21.1

Model Card Authors

ALMAnaCH team, Inria Paris

Additional Resources

• 🔗 GitHub : https://github.com/NathanGodey/gapetron
• 📄 Paper : 📄 Paper Link
• 🔧 Evaluation Tools : https://gitlab.inria.fr/almanach/lm-evaluation-harness-gaperon
• 📊 Base Model : almanach/Gaperon-1125-8B

Citation

If you use this model, please cite:

@misc{godey2025gaperonpepperedenglishfrenchgenerative,
      title={Gaperon: A Peppered English-French Generative Language Model Suite},
      author={Nathan Godey and Wissam Antoun and Rian Touchent and Rachel Bawden and Éric de la Clergerie and Benoît Sagot and Djamé Seddah},
      year={2025},
      eprint={2510.25771},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.25771},
}

Acknowledgments

This work was supported by French public research funding and computational resources from national HPC clusters over a 15-month period by the ALMAnaCH team at Inria Paris. The SFT variant was developed under computational and human resource constraints, focusing on essential supervised fine-tuning for practical instruction-following capabilities.