Akicou/GLM-4.7-Flash-REAP-39
LoRA-style (fewer layers) vs zai-org/GLM-4.7-Flash. Akicou/GLM-4.7-Flash-REAP-39 matches part of zai-org/GLM-4.7-Flash (5949 of its 9703 tensors; the rest of zai-org/GLM-4.7-Flash 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 46 of 47 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/189, MLP 1317/1504, Attention 0/235, lm_head 0/1
3 derivatives checked
- mradermacher/GLM-4.7-Flash-REAP-39-i1-GGUF: Quantization
- mradermacher/GLM-4.7-Flash-REAP-39-GGUF: Quantization
- Akicou/GLM-4.7-Flash-REAP-39-GGUF: Quantization