Blackfrost-AI/CYBER-FROST-3.8-BF16
Partial fine-tune vs Qwen/Qwen3.8-Flash-Next. Partial change: 69% of the weights changed (attention, lm_head, mlp, other), in 48 of 48 layers.
Tensors changed: Norms 0/302, other 4/8, Attention 372/608, MLP 432/602, Embeddings 0/137, lm_head 1/1
7 derivatives checked
- freakyskittle/CYBER-FROST-3.8-GGUF: Not comparable yet
- peasantsmith/CYBER-FROST-3.8-PS-GUFF: Not comparable yet
- 0xSojalSec/CYBER-FROST-3.8-GGUF: Not comparable yet
- vcruz305/CYBER-FROST-3.8-EXL3-SAGE-3.87bpw: Same weights, converted
- Blackfrost-AI/CYBER-FROST-3.8-NVFP4-V2: Quantization
- Kazuya-Nomura-zer0/CYBER-FROST-3.8-GGUF: Not comparable yet
- Blackfrost-AI/CYBER-FROST-3.8-FP8: Quantization
1 other repos hold byte-for-byte identical weight files.