fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit

<p align="center"> <img src="assets/banner_4bit.png" alt="gpt-oss-20b · Hermes-Agent tool finetune · MXFP4" width="100%" /> </p>

gpt-oss-20b · Hermes-Agent tool finetune · MXFP4

4-bit MXFP4 quantization of the BF16 checkpoint. Fits in 16 GB of VRAM, runs on a Colab T4, keeps the full 8k context.

• —Format — Safetensors, MXFP4
• —Size on disk — ~12 GB
• —VRAM (inference) — ~14-16 GB
• —Recommended runtime — Unsloth, Transformers + bitsandbytes

What this is

A tool-use finetune of OpenAI's gpt-oss-20b for Hermes-Agent, a local agent framework that needs models which call tools reliably, follow multi-turn instructions, and don't argue with system prompts.

The base model is the 21B-parameter (3.6B active) Mixture-of-Experts release from OpenAI. This finetune preserves the Harmony chat template and the reasoning-effort knob, and improves:

• —Function-calling adherence (correct JSON, no commentary mid-call)
• —Long agent loops (10+ turns of tool → observe → plan)
• —System-prompt fidelity (respects role boundaries and refusal/allow-list rules)

It is not affiliated with NousResearch's Hermes model series. "Hermes-Agent" here refers to the local agent framework only.

Quickstart

Unsloth (recommended)

from unsloth import FastLanguageModel

model, tok = FastLanguageModel.from_pretrained(
    "fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit",
    max_seq_length=8192,
    load_in_4bit=True,
)
FastLanguageModel.for_inference(model)

Transformers

from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
import torch

tok = AutoTokenizer.from_pretrained("fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit")
model = AutoModelForCausalLM.from_pretrained(
    "fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit",
    quantization_config=BitsAndBytesConfig(load_in_4bit=True,
        bnb_4bit_compute_dtype=torch.bfloat16,
        bnb_4bit_quant_type="nf4"),
    device_map="auto",
)

Hermes-Agent integration

Add a profile in ~/.hermes/config.yaml:

profiles:
  gpt-oss-20b-tools:
    provider: openai
    base_url: http://127.0.0.1:1234/v1   # LM Studio / vLLM / mlx_lm.server
    model: fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit
    temperature: 0.7
    top_p: 0.95
    min_p: 0.1                            # important for MoE stability
    max_tokens: 8192
    tool_choice: auto

Then hermes profile use gpt-oss-20b-tools and the agent loop will route tool calls through this model.

Sampling

Harmony reasoning effort: set the system message to Reasoning: low|medium|high. high is roughly 3-4x more output tokens but noticeably better on multi-step tool plans.

Training

• —Base: openai/gpt-oss-20b
• —Method: LoRA SFT (rank 64, alpha 16) merged back into BF16
• —Frame: Unsloth + TRL on a single H100 (80 GB)
• —Data: ~42k tool-use traces from Hermes-Agent sessions, filtered for successful tool calls and clean JSON. No synthetic distillation.
• —Length: 8192 tokens, packing on
• —Loss: assistant-only, mask user/system/tool

The _16bit repo holds the merged BF16 weights. The _4bit, _mlx, and _gguf repos are quantizations of that checkpoint.

Limitations

• —Math and code-generation are unchanged from the base — this finetune optimizes the agent loop, not raw reasoning.
• —The model can over-call tools when given vague instructions. Add a "if you can answer directly, do so" line to the system prompt.
• —English only. Other languages were not in the training mix.
• —Not safety-tuned beyond what gpt-oss-20b already provides.

Other formats

• —BF16 reference — full precision, vLLM / Transformers
• —MXFP4 4-bit — fits a 16 GB GPU
• —MLX — Apple Silicon native
• —GGUF — llama.cpp / Ollama / LM Studio

License

Apache-2.0, inherited from the base model. No additional restrictions.

Citation

@misc{fesalfayed_gptoss20b_hermesagent_2025,
  author = {Fayed, Fesal},
  title  = {gpt-oss-20b Hermes-Agent tool finetune (4bit)},
  year   = {2025},
  url    = {https://huggingface.co/fesalfayed/gpt-oss-20b-hermes_agent-tool-finetune_4bit},
}