halley-ai/gpt-oss-120b-MLX-bf16

gpt-oss-120b — MLX bf16 (non-quantized)

Summary. This is a non-quantized MLX conversion of gpt-oss-120B in bfloat16 (bf16). Built for Apple Silicon with Metal acceleration.

• Base model: openai/gpt-oss-120b (Apache-2.0)
• Precision: bfloat16 (no quantization)
• Files: MLX weight shards + config.json; tokenizer files included for drop-in use
• Intended use: local inference / research on M-series Macs
• Not intended for: safety-critical decisions; outputs may be inaccurate or biased

Requirements

Runs on Apple Silicon (M1 or newer) with macOS ≥ 13.5 via MLX (Metal).

• Not supported: Intel macOS / Linux / Windows (consider a GGUF build + llama.cpp instead).
• Memory guidance: large unified memory recommended (e.g., 64–96 GB). The effective GPU working set is capped by Metal’s budget; keep 5–10% headroom.

How to use (MLX)

pip install mlx-lm

# Python API (uses tokenizer bundled with this repo)
from mlx_lm import load, generate

model, tokenizer = load("halley-ai/gpt-oss-120b-MLX-bf16")
print(generate(
    model, tokenizer,
    prompt="Explain the Chudnovsky algorithm to compute π.",
    max_tokens=256, max_kv_size=512
))

# CLI
python -m mlx_lm generate --model halley-ai/gpt-oss-120b-MLX-bf16 \
  --prompt "Explain the Chudnovsky algorithm to compute pi." \
  --max-kv-size 512 --max-tokens 256

Evaluation

Perplexity (PPL) streaming evaluation on WikiText-2 (raw, test); fast preset with window=stride=4096, ~100k tokens, EOS inserted between docs.

Variant	PPL (ctx=4096, fast)
MLX bf16 (non-quant)	7.38
MLX 8-bit (gs=32)	7.39
MLX 6-bit (gs=64)	7.40

Notes

• Results from local runs on Apple Silicon using MLX; numbers vary slightly with tokenizer details, logits dtype, and token subset.
• For more sensitive comparisons, use overlapping windows (e.g., --stride 512) and evaluate the full split.

Conversion details (provenance)

python -m mlx_lm convert \
  --hf-path openai/gpt-oss-120b \
  --mlx-path gpt-oss-120b-MLX-bf16 \
  --dtype bfloat16

Sibling & reference models

• halley-ai/gpt-oss-120b-MLX-8bit-gs32 (int8, group size 32)
• halley-ai/gpt-oss-120b-MLX-6bit-gs64 (int6, group size 64)

Limitations & biases

Outputs may be factually wrong or unsafe. Do not use for medical, legal, or financial decisions without human review. Large models can be sensitive to prompts; prefer explicit instructions and structure.

License & credits

• License: Apache-2.0 (inherits from base model)
• Base model: OpenAI gpt-oss-120B
• Conversion: Halley AI Lab (MLX bf16)
• Please cite both the base model and this repository when you use the weights.