EpistemeAI/Hercules-Coder-E4B-it

This finetuned model is specialized in STEM like LCB, CodeForce, AIME24, AIME25, AMC23, MATH500.

Note:

• Currently only text is supported.
• Ollama: ollama run hf.co/unsloth/gemma-3n-E4B-it-GGUF:Q4_K_XL - auto-sets correct chat template and settings
• Set temperature = 1.0, top_k = 64, top_p = 0.95, min_p = 0.0
• Gemma 3n max tokens (context length): 32K. Gemma 3n chat template:

Use unsloth inference

!pip install --upgrade transformers

import torch
from transformers import pipeline
model_id = "EpistemeAI/Hercules-Coder-E4B-it"
pipe = pipeline(
    "text-generation", 
    model=model_id, 
    torch_dtype=torch.bfloat16, 
    device_map="auto"
)
print(pipe("Write me a Python function to calculate the nth fibonacci number."))

Benchmark results (5 shot):

Tasks	Version	Filter	n-shot	Metric		Value
arc_challenge	1	none	5	acc	↑	0.5759
hellaswag	1	none	5	acc	↑	0.7651
winogrande	1	none	5	acc	↑	0.7526

GPQA Diamond result

Tasks	Version	Filter	n-shot	Metric		Value
gpqa_diamond_zeroshot	1	none	0	acc	↑	0.2516
		none	0	acc_norm	↑	0.2516

Uploaded finetuned model

• Developed by: EpistemeAI
• License: apache-2.0
• Finetuned from model : unsloth/gemma-3n-e4b-unsloth-bnb-4bit

This gemma3n model was trained 2x faster with Unsloth and Huggingface's TRL library.

Citations

@misc{liu2025rstarcoderscalingcompetitivecode,
      title={rStar-Coder: Scaling Competitive Code Reasoning with a Large-Scale Verified Dataset}, 
      author={Yifei Liu and Li Lyna Zhang and Yi Zhu and Bingcheng Dong and Xudong Zhou and Ning Shang and Fan Yang and Mao Yang},
      year={2025},
      eprint={2505.21297},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.21297}, 
}