LLaMAX/GlotMAX-17-8B
Partial fine-tune vs Qwen/Qwen3-8B. Partial change: 45% of the weights changed (attention, mlp, norms), in 19 of 36 layers.
Tensors changed: Embeddings 0/1, Norms 70/145, MLP 57/108, Attention 76/144, lm_head 0/1
6 derivatives checked
- mradermacher/Qwen3-XPlus-17langs-8B-i1-GGUF: Quantization
- mradermacher/GlotMAX-17-8B-i1-GGUF: Quantization
- mradermacher/Qwen3-XPlus-8B-i1-GGUF: Quantization
- mradermacher/GlotMAX-17-8B-GGUF: Same weights, converted
- mradermacher/Qwen3-XPlus-17langs-8B-GGUF: Same weights, converted
- mradermacher/Qwen3-XPlus-8B-GGUF: Same weights, converted