GGUF v3: 720 tensor descriptors; payloads not read
gemma4.block_count = 42
gemma4.embedding_length = 2560
gemma4.feed_forward_length = 10240
gemma4.embedding_length_per_layer_input = 256
layer-0 FFN and PLE descriptors:
blk.0.ffn_gate.weight: IQ4_XS ne=[2560, 10240] bytes=13926400
blk.0.ffn_up.weight: IQ4_XS ne=[2560, 10240] bytes=13926400
blk.0.ffn_down.weight: IQ4_XS ne=[10240, 2560] bytes=13926400
per_layer_token_embd.weight: Q5_K ne=[10752, 262144] bytes=1937768448
per_layer_model_proj.weight: BF16 ne=[2560, 10752] bytes=55050240
per_layer_proj_norm.weight: F32 ne=[256] bytes=1024
blk.0.inp_gate.weight: F32 ne=[2560, 256] bytes=2621440
blk.0.proj.weight: F32 ne=[256, 2560] bytes=2621440
