openai/gpt-oss-safeguard-20b
Partial fine-tune vs openai/gpt-oss-20b. Partial change: 69% of the weights changed (attention, embeddings, lm_head, mlp, norms), in 24 of 24 layers.
Tensors changed: Norms 49/49, MLP 125/192, Attention 174/216, lm_head 1/1, Embeddings 1/1
9 derivatives checked
- lmstudio-community/gpt-oss-safeguard-20b-MLX-MXFP4: Quantization
- unsloth/gpt-oss-safeguard-20b-GGUF: Same weights, converted
- lmstudio-community/gpt-oss-safeguard-20b-GGUF: Quantization
- mradermacher/gpt-oss-safeguard-20b-i1-GGUF: Quantization
- mradermacher/gpt-oss-safeguard-20b-GGUF: Quantization
- unsloth/gpt-oss-safeguard-20b: Identical copy
- DevQuasar/openai.gpt-oss-safeguard-20b-GGUF: Same weights, converted
- unsloth/gpt-oss-safeguard-20b-BF16: Same weights, repacked
- SEOKDONG/gpt-oss-safeguard-20b-kor-enterprise: LoRA-style
5 other repos hold byte-for-byte identical weight files.