cerebras/GLM-4.7-REAP-218B-A32B
LoRA-style (fewer layers) vs zai-org/GLM-4.7. cerebras/GLM-4.7-REAP-218B-A32B matches part of zai-org/GLM-4.7 (26923 of its 44691 tensors; the rest of zai-org/GLM-4.7 has no counterpart). On the shared part: Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, up_proj changed, in 89 of 92 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/369, MLP 859/1175, Attention 0/644, lm_head 0/1
5 derivatives checked
- unsloth/GLM-4.7-REAP-218B-A32B-GGUF: Same weights, converted
- bartowski/cerebras_GLM-4.7-REAP-218B-A32B-GGUF: Quantization
- mradermacher/GLM-4.7-REAP-218B-A32B-i1-GGUF: Quantization
- mradermacher/GLM-4.7-REAP-218B-A32B-GGUF: Quantization
- 0xSero/GLM-4.7-218B-W4A16: Quantization
1 other repos hold byte-for-byte identical weight files.