ukisai/Swift-1.5-Qwen3.8-27b
Partial fine-tune vs Qwen/Qwen3.8-27B. Partial change: 88% of the weights changed (attention, mlp, norms), in 64 of 64 layers.
Tensors changed: Norms 119/325, Attention 353/560, MLP 192/307, Embeddings 0/5, lm_head 0/1, other 0/1
48 derivatives checked
- ukisai/Swift-1.5-Qwen3.8-27B-GSQ-RCO-GGUF: Quantization
- ukisai/Swift-1.5-Qwen3.8-27B-GGUF: Quantization
- bartowski/ukisai_Swift-1.5-Qwen3.8-27b-GGUF: Same weights, converted
- ukisai/Swift-1.5-Qwen3.8-27b-INT4: Quantization
- jegreene/Swift-1.5-Qwen3.8-27B-heretic-GSQ-RCO-GGUF: Quantization
- ukisai/Swift-1.5-Qwen3.8-27b-W4A16-AWQ: Quantization
- ukisai/Swift-1.5-Qwen3.8-27b-NVFP4: Quantization
- tsaipifong/Swift-1.5-Qwen3.8-27b-MXFP4-GGUF: Quantization
- ukisai/Swift-1.5-4bit-MLX: Quantization
- mradermacher/Swift-1.5-Qwen3.8-27b-i1-GGUF: Quantization
- ukisai/Swift-1.5-5bit-MLX: Quantization
- gray117/Swift-1.5-Qwen3.8-27B-Q4_0_ROCMFP4_STRIX: Quantization
- cyankiwi/Swift-1.5-Qwen3.8-27B-AWQ-INT4: Quantization
- poisonxa/Swift-1.5-Qwen3.8-27B-PXQN-OneCard: Full fine-tune
- ukisai/Swift-1.5-Qwen3.8-27b-W4A16-AutoRound: Quantization
- mradermacher/Swift-1.5-Qwen3.8-27b-GGUF: Quantization
- dicksondickson/Swift-1.5-Qwen3.8-27b-oQ5e-fp16-mtp: Quantization
- cyankiwi/Swift-1.5-Qwen3.8-27B-AWQ-BF16-INT4: Quantization
- ukisai/Swift-1.5-3bit-MLX-TextOnly: Quantization
- dicksondickson/Swift-1.5-Qwen3.8-27b-oQ8e-fp16-mtp: Quantization
- realderpz/Swift-1.5-Qwen3.8-27B-FP8: Quantization
- yottle/Swift-1.5-Qwen3.8-27b-oQ4e-mtp: Quantization
- bjonor/Swift-1.5-Qwen3.8-27B-GPTQ-Int4-sym-G128-MTP-BF16: Quantization
- ajgazin/Swift-1.5-Qwen3.8-27B-Uncensored-MTP: Abliterated
- kunthawat/swift-1.5-qwen3.8-27b-with-unsloth-iq3-xxs-gguf: Quantization
- daavidhauser/Swift-1.5-Qwen3.8-27B-W4A16-HyperQwen: Quantization
- ukisai/Swift-1.5-Qwen3.8-27b-Quark-FP8-dynamic-AMD: Quantization
- jordi123/Swift-1.5-Qwen3.8-27b-i1-GGUF: Quantization
- CleverBadger/Swift-1.5-Qwen3.8-27B-Badger-Mixed-GGUF: Quantization
- dicksondickson/Swift-1.5-Qwen3.8-27b-oQ6e-bf16-mtp-MLX: Quantization
1 other repos hold byte-for-byte identical weight files.