dhiltgen/ gguf-transfer-test:qwen38-27b-bf16-4g-20261003

4 2 weeks ago

vision
... /
model
5a1e1b962aab ยท 3.4GB
    Metadata
  • split.count
    15
  • split.no
    14
  • split.tensors.count
    866
    • Tensor
    blk.64
  • blk.64.attn_k.weight
    BF16
    [5120, 1024]
  • blk.64.attn_k_norm.weight
    F32
    [256]
  • blk.64.attn_norm.weight
    F32
    [5120]
  • blk.64.attn_output.weight
    BF16
    [6144, 5120]
  • blk.64.attn_q.weight
    BF16
    [5120, 12288]
  • blk.64.attn_q_norm.weight
    F32
    [256]
  • blk.64.attn_v.weight
    BF16
    [5120, 1024]
  • blk.64.ffn_down.weight
    BF16
    [17408, 5120]
  • blk.64.ffn_gate.weight
    BF16
    [5120, 17408]
  • blk.64.ffn_up.weight
    BF16
    [5120, 17408]
  • blk.64.nextn.eh_proj.weight
    BF16
    [10240, 5120]
  • blk.64.nextn.enorm.weight
    F32
    [5120]
  • blk.64.nextn.hnorm.weight
    F32
    [5120]
  • blk.64.nextn.shared_head_norm.weight
    F32
    [5120]
  • blk.64.post_attention_norm.weight
    F32
    [5120]
  • output.weight
    BF16
    [5120, 248320]