na0x2c6/ELYZA-Thinking-1.0-llm-jp-4-32b-a3b-GGUF

Same weights, converted vs elyza/ELYZA-Thinking-1.0-llm-jp-4-32b-a3b. Same weights stored in another dtype (BF16 → BF16/F32), plus 96 new tensors.

Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/129, MLP 0/32, Attention 0/128

On the Hub · Downloads on Model Pulse