unsloth/gpt-oss-20b-BF16
Same weights, repacked vs openai/gpt-oss-20b. Same weights in different files (re-saved, re-sharded or new metadata), plus 48 new tensors.
Tensors changed: Norms 0/49, MLP 0/96, Attention 0/216, lm_head 0/1, Embeddings 0/1
9 derivatives checked
- huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated: Full fine-tune
- microsoft/OptiMind-SFT: Full fine-tune
- amd/gpt-oss-20b-BF16-w8a8-llmcompressor: Quantization
- Melikshah/gpt-oss-20b-clinical-cot-gguf: LoRA-style
- amd/gpt-oss-20b-BF16-w4a16-llmcompressor: Quantization
- win10/GPT-OSS-26B-abliterated-Preview: Partial fine-tune
- win10/GPT-OSS-30B-Preview: Partial fine-tune (fewer layers)
- surfiniaburger/dipg-safety-agent-v2-float16: LoRA-style
- Kwoya/Spyra-20B-v.1.1: Full fine-tune
18 other repos hold byte-for-byte identical weight files.