70 Pulls Updated 6 months ago

630dad89c4f4 · 2.9GB
    Metadata
  • general.architecture
    bloom
  • general.file_type
    F16
  • bloom.attention.head_count
    16
  • bloom.attention.head_count_kv
    16
  • bloom.attention.layer_norm_epsilon
    1e-05
  • bloom.block_count
    24
  • bloom.context_length
    1536
  • bloom.embedding_length
    1536
  • bloom.feed_forward_length
    6144
  • tokenizer.ggml.bos_token_id
    1
  • tokenizer.ggml.eos_token_id
    2
  • tokenizer.ggml.merges
    [à ¤ à ¦ Ġ à à ¥ Ġ Ġ ...]
  • tokenizer.ggml.model
    gpt2
  • tokenizer.ggml.padding_token_id
    3
  • tokenizer.ggml.token_type
    [3 3 3 3 1 ...]
  • tokenizer.ggml.tokens
    [<unk> <s> </s> <pad> ! ...]
  • tokenizer.ggml.unknown_token_id
    0
  • Tensors
  • Name
    Type
    Shape
  • token_embd.weight
    F16
    [1536 250880]
  • token_embd_norm.bias
    F32
    [1536]
  • token_embd_norm.weight
    F32
    [1536]
  • blk.0
  • blk.0.attn_norm.bias
    F32
    [1536]
  • blk.0.attn_norm.weight
    F32
    [1536]
  • blk.0.ffn_down.bias
    F32
    [1536]
  • blk.0.ffn_down.weight
    F16
    [6144 1536]
  • blk.0.ffn_up.bias
    F32
    [6144]
  • blk.0.ffn_up.weight
    F16
    [1536 6144]
  • blk.0.ffn_norm.bias
    F32
    [1536]
  • blk.0.ffn_norm.weight
    F32
    [1536]
  • blk.0.attn_output.bias
    F32
    [1536]
  • blk.0.attn_output.weight
    F16
    [1536 1536]
  • blk.0.attn_qkv.bias
    F32
    [4608]
  • blk.0.attn_qkv.weight
    F16
    [1536 4608]
  • blk.1
  • blk.1.attn_norm.bias
    F32
    [1536]
  • blk.1.attn_norm.weight
    F32
    [1536]
  • blk.1.ffn_down.bias
    F32
    [1536]
  • blk.1.ffn_down.weight
    F16
    [6144 1536]
  • blk.1.ffn_up.bias
    F32
    [6144]
  • blk.1.ffn_up.weight
    F16
    [1536 6144]
  • blk.1.ffn_norm.bias
    F32
    [1536]
  • blk.1.ffn_norm.weight
    F32
    [1536]
  • blk.1.attn_output.bias
    F32
    [1536]
  • blk.1.attn_output.weight
    F16
    [1536 1536]
  • blk.1.attn_qkv.bias
    F32
    [4608]
  • blk.1.attn_qkv.weight
    F16
    [1536 4608]
  • blk.2
  • blk.2.attn_norm.bias
    F32
    [1536]
  • blk.2.attn_norm.weight
    F32
    [1536]
  • blk.2.ffn_down.bias
    F32
    [1536]
  • blk.2.ffn_down.weight
    F16
    [6144 1536]
  • blk.2.ffn_up.bias
    F32
    [6144]
  • blk.2.ffn_up.weight
    F16
    [1536 6144]
  • blk.2.ffn_norm.bias
    F32
    [1536]
  • blk.2.ffn_norm.weight
    F32
    [1536]
  • blk.2.attn_output.bias
    F32
    [1536]
  • blk.2.attn_output.weight
    F16
    [1536 1536]
  • blk.2.attn_qkv.bias
    F32
    [4608]
  • blk.2.attn_qkv.weight
    F16
    [1536 4608]
  • blk.3
  • blk.3.attn_norm.bias
    F32
    [1536]
  • blk.3.attn_norm.weight
    F32
    [1536]
  • blk.3.ffn_down.bias
    F32
    [1536]
  • blk.3.ffn_down.weight
    F16
    [6144 1536]
  • blk.3.ffn_up.bias
    F32
    [6144]
  • blk.3.ffn_up.weight
    F16
    [1536 6144]
  • blk.3.ffn_norm.bias
    F32
    [1536]
  • blk.3.ffn_norm.weight
    F32
    [1536]
  • blk.3.attn_output.bias
    F32
    [1536]
  • blk.3.attn_output.weight
    F16
    [1536 1536]
  • blk.3.attn_qkv.bias
    F32
    [4608]
  • blk.3.attn_qkv.weight
    F16
    [1536 4608]
  • blk.4
  • blk.4.attn_norm.bias
    F32
    [1536]
  • blk.4.attn_norm.weight
    F32
    [1536]
  • blk.4.ffn_down.bias
    F32
    [1536]
  • blk.4.ffn_down.weight
    F16
    [6144 1536]
  • blk.4.ffn_up.bias
    F32
    [6144]
  • blk.4.ffn_up.weight
    F16
    [1536 6144]
  • blk.4.ffn_norm.bias
    F32
    [1536]
  • blk.4.ffn_norm.weight
    F32
    [1536]
  • blk.4.attn_output.bias
    F32
    [1536]
  • blk.4.attn_output.weight
    F16
    [1536 1536]
  • blk.4.attn_qkv.bias
    F32
    [4608]
  • blk.4.attn_qkv.weight
    F16
    [1536 4608]
  • blk.5
  • blk.5.attn_norm.bias
    F32
    [1536]
  • blk.5.attn_norm.weight
    F32
    [1536]
  • blk.5.ffn_down.bias
    F32
    [1536]
  • blk.5.ffn_down.weight
    F16
    [6144 1536]
  • blk.5.ffn_up.bias
    F32
    [6144]
  • blk.5.ffn_up.weight
    F16
    [1536 6144]
  • blk.5.ffn_norm.bias
    F32
    [1536]
  • blk.5.ffn_norm.weight
    F32
    [1536]
  • blk.5.attn_output.bias
    F32
    [1536]
  • blk.5.attn_output.weight
    F16
    [1536 1536]
  • blk.5.attn_qkv.bias
    F32
    [4608]
  • blk.5.attn_qkv.weight
    F16
    [1536 4608]
  • blk.6
  • blk.6.attn_norm.bias
    F32
    [1536]
  • blk.6.attn_norm.weight
    F32
    [1536]
  • blk.6.ffn_down.bias
    F32
    [1536]
  • blk.6.ffn_down.weight
    F16
    [6144 1536]
  • blk.6.ffn_up.bias
    F32
    [6144]
  • blk.6.ffn_up.weight
    F16
    [1536 6144]
  • blk.6.ffn_norm.bias
    F32
    [1536]
  • blk.6.ffn_norm.weight
    F32
    [1536]
  • blk.6.attn_output.bias
    F32
    [1536]
  • blk.6.attn_output.weight
    F16
    [1536 1536]
  • blk.6.attn_qkv.bias
    F32
    [4608]
  • blk.6.attn_qkv.weight
    F16
    [1536 4608]
  • blk.7
  • blk.7.attn_norm.bias
    F32
    [1536]
  • blk.7.attn_norm.weight
    F32
    [1536]
  • blk.7.ffn_down.bias
    F32
    [1536]
  • blk.7.ffn_down.weight
    F16
    [6144 1536]
  • blk.7.ffn_up.bias
    F32
    [6144]
  • blk.7.ffn_up.weight
    F16
    [1536 6144]
  • blk.7.ffn_norm.bias
    F32
    [1536]
  • blk.7.ffn_norm.weight
    F32
    [1536]
  • blk.7.attn_output.bias
    F32
    [1536]
  • blk.7.attn_output.weight
    F16
    [1536 1536]
  • blk.7.attn_qkv.bias
    F32
    [4608]
  • blk.7.attn_qkv.weight
    F16
    [1536 4608]
  • blk.8
  • blk.8.attn_norm.bias
    F32
    [1536]
  • blk.8.attn_norm.weight
    F32
    [1536]
  • blk.8.ffn_down.bias
    F32
    [1536]
  • blk.8.ffn_down.weight
    F16
    [6144 1536]
  • blk.8.ffn_up.bias
    F32
    [6144]
  • blk.8.ffn_up.weight
    F16
    [1536 6144]
  • blk.8.ffn_norm.bias
    F32
    [1536]
  • blk.8.ffn_norm.weight
    F32
    [1536]
  • blk.8.attn_output.bias
    F32
    [1536]
  • blk.8.attn_output.weight
    F16
    [1536 1536]
  • blk.8.attn_qkv.bias
    F32
    [4608]
  • blk.8.attn_qkv.weight
    F16
    [1536 4608]
  • blk.9
  • blk.9.attn_norm.bias
    F32
    [1536]
  • blk.9.attn_norm.weight
    F32
    [1536]
  • blk.9.ffn_down.bias
    F32
    [1536]
  • blk.9.ffn_down.weight
    F16
    [6144 1536]
  • blk.9.ffn_up.bias
    F32
    [6144]
  • blk.9.ffn_up.weight
    F16
    [1536 6144]
  • blk.9.ffn_norm.bias
    F32
    [1536]
  • blk.9.ffn_norm.weight
    F32
    [1536]
  • blk.9.attn_output.bias
    F32
    [1536]
  • blk.9.attn_output.weight
    F16
    [1536 1536]
  • blk.9.attn_qkv.bias
    F32
    [4608]
  • blk.9.attn_qkv.weight
    F16
    [1536 4608]
  • blk.10
  • blk.10.attn_norm.bias
    F32
    [1536]
  • blk.10.attn_norm.weight
    F32
    [1536]
  • blk.10.ffn_down.bias
    F32
    [1536]
  • blk.10.ffn_down.weight
    F16
    [6144 1536]
  • blk.10.ffn_up.bias
    F32
    [6144]
  • blk.10.ffn_up.weight
    F16
    [1536 6144]
  • blk.10.ffn_norm.bias
    F32
    [1536]
  • blk.10.ffn_norm.weight
    F32
    [1536]
  • blk.10.attn_output.bias
    F32
    [1536]
  • blk.10.attn_output.weight
    F16
    [1536 1536]
  • blk.10.attn_qkv.bias
    F32
    [4608]
  • blk.10.attn_qkv.weight
    F16
    [1536 4608]
  • blk.11
  • blk.11.attn_norm.bias
    F32
    [1536]
  • blk.11.attn_norm.weight
    F32
    [1536]
  • blk.11.ffn_down.bias
    F32
    [1536]
  • blk.11.ffn_down.weight
    F16
    [6144 1536]
  • blk.11.ffn_up.bias
    F32
    [6144]
  • blk.11.ffn_up.weight
    F16
    [1536 6144]
  • blk.11.ffn_norm.bias
    F32
    [1536]
  • blk.11.ffn_norm.weight
    F32
    [1536]
  • blk.11.attn_output.bias
    F32
    [1536]
  • blk.11.attn_output.weight
    F16
    [1536 1536]
  • blk.11.attn_qkv.bias
    F32
    [4608]
  • blk.11.attn_qkv.weight
    F16
    [1536 4608]
  • blk.12
  • blk.12.attn_norm.bias
    F32
    [1536]
  • blk.12.attn_norm.weight
    F32
    [1536]
  • blk.12.ffn_down.bias
    F32
    [1536]
  • blk.12.ffn_down.weight
    F16
    [6144 1536]
  • blk.12.ffn_up.bias
    F32
    [6144]
  • blk.12.ffn_up.weight
    F16
    [1536 6144]
  • blk.12.ffn_norm.bias
    F32
    [1536]
  • blk.12.ffn_norm.weight
    F32
    [1536]
  • blk.12.attn_output.bias
    F32
    [1536]
  • blk.12.attn_output.weight
    F16
    [1536 1536]
  • blk.12.attn_qkv.bias
    F32
    [4608]
  • blk.12.attn_qkv.weight
    F16
    [1536 4608]
  • blk.13
  • blk.13.attn_norm.bias
    F32
    [1536]
  • blk.13.attn_norm.weight
    F32
    [1536]
  • blk.13.ffn_down.bias
    F32
    [1536]
  • blk.13.ffn_down.weight
    F16
    [6144 1536]
  • blk.13.ffn_up.bias
    F32
    [6144]
  • blk.13.ffn_up.weight
    F16
    [1536 6144]
  • blk.13.ffn_norm.bias
    F32
    [1536]
  • blk.13.ffn_norm.weight
    F32
    [1536]
  • blk.13.attn_output.bias
    F32
    [1536]
  • blk.13.attn_output.weight
    F16
    [1536 1536]
  • blk.13.attn_qkv.bias
    F32
    [4608]
  • blk.13.attn_qkv.weight
    F16
    [1536 4608]
  • blk.14
  • blk.14.attn_norm.bias
    F32
    [1536]
  • blk.14.attn_norm.weight
    F32
    [1536]
  • blk.14.ffn_down.bias
    F32
    [1536]
  • blk.14.ffn_down.weight
    F16
    [6144 1536]
  • blk.14.ffn_up.bias
    F32
    [6144]
  • blk.14.ffn_up.weight
    F16
    [1536 6144]
  • blk.14.ffn_norm.bias
    F32
    [1536]
  • blk.14.ffn_norm.weight
    F32
    [1536]
  • blk.14.attn_output.bias
    F32
    [1536]
  • blk.14.attn_output.weight
    F16
    [1536 1536]
  • blk.14.attn_qkv.bias
    F32
    [4608]
  • blk.14.attn_qkv.weight
    F16
    [1536 4608]
  • blk.15
  • blk.15.attn_norm.bias
    F32
    [1536]
  • blk.15.attn_norm.weight
    F32
    [1536]
  • blk.15.ffn_down.bias
    F32
    [1536]
  • blk.15.ffn_down.weight
    F16
    [6144 1536]
  • blk.15.ffn_up.bias
    F32
    [6144]
  • blk.15.ffn_up.weight
    F16
    [1536 6144]
  • blk.15.ffn_norm.bias
    F32
    [1536]
  • blk.15.ffn_norm.weight
    F32
    [1536]
  • blk.15.attn_output.bias
    F32
    [1536]
  • blk.15.attn_output.weight
    F16
    [1536 1536]
  • blk.15.attn_qkv.bias
    F32
    [4608]
  • blk.15.attn_qkv.weight
    F16
    [1536 4608]
  • blk.16
  • blk.16.attn_norm.bias
    F32
    [1536]
  • blk.16.attn_norm.weight
    F32
    [1536]
  • blk.16.ffn_down.bias
    F32
    [1536]
  • blk.16.ffn_down.weight
    F16
    [6144 1536]
  • blk.16.ffn_up.bias
    F32
    [6144]
  • blk.16.ffn_up.weight
    F16
    [1536 6144]
  • blk.16.ffn_norm.bias
    F32
    [1536]
  • blk.16.ffn_norm.weight
    F32
    [1536]
  • blk.16.attn_output.bias
    F32
    [1536]
  • blk.16.attn_output.weight
    F16
    [1536 1536]
  • blk.16.attn_qkv.bias
    F32
    [4608]
  • blk.16.attn_qkv.weight
    F16
    [1536 4608]
  • blk.17
  • blk.17.attn_norm.bias
    F32
    [1536]
  • blk.17.attn_norm.weight
    F32
    [1536]
  • blk.17.ffn_down.bias
    F32
    [1536]
  • blk.17.ffn_down.weight
    F16
    [6144 1536]
  • blk.17.ffn_up.bias
    F32
    [6144]
  • blk.17.ffn_up.weight
    F16
    [1536 6144]
  • blk.17.ffn_norm.bias
    F32
    [1536]
  • blk.17.ffn_norm.weight
    F32
    [1536]
  • blk.17.attn_output.bias
    F32
    [1536]
  • blk.17.attn_output.weight
    F16
    [1536 1536]
  • blk.17.attn_qkv.bias
    F32
    [4608]
  • blk.17.attn_qkv.weight
    F16
    [1536 4608]
  • blk.18
  • blk.18.attn_norm.bias
    F32
    [1536]
  • blk.18.attn_norm.weight
    F32
    [1536]
  • blk.18.ffn_down.bias
    F32
    [1536]
  • blk.18.ffn_down.weight
    F16
    [6144 1536]
  • blk.18.ffn_up.bias
    F32
    [6144]
  • blk.18.ffn_up.weight
    F16
    [1536 6144]
  • blk.18.ffn_norm.bias
    F32
    [1536]
  • blk.18.ffn_norm.weight
    F32
    [1536]
  • blk.18.attn_output.bias
    F32
    [1536]
  • blk.18.attn_output.weight
    F16
    [1536 1536]
  • blk.18.attn_qkv.bias
    F32
    [4608]
  • blk.18.attn_qkv.weight
    F16
    [1536 4608]
  • blk.19
  • blk.19.attn_norm.bias
    F32
    [1536]
  • blk.19.attn_norm.weight
    F32
    [1536]
  • blk.19.ffn_down.bias
    F32
    [1536]
  • blk.19.ffn_down.weight
    F16
    [6144 1536]
  • blk.19.ffn_up.bias
    F32
    [6144]
  • blk.19.ffn_up.weight
    F16
    [1536 6144]
  • blk.19.ffn_norm.bias
    F32
    [1536]
  • blk.19.ffn_norm.weight
    F32
    [1536]
  • blk.19.attn_output.bias
    F32
    [1536]
  • blk.19.attn_output.weight
    F16
    [1536 1536]
  • blk.19.attn_qkv.bias
    F32
    [4608]
  • blk.19.attn_qkv.weight
    F16
    [1536 4608]
  • blk.20
  • blk.20.attn_norm.bias
    F32
    [1536]
  • blk.20.attn_norm.weight
    F32
    [1536]
  • blk.20.ffn_down.bias
    F32
    [1536]
  • blk.20.ffn_down.weight
    F16
    [6144 1536]
  • blk.20.ffn_up.bias
    F32
    [6144]
  • blk.20.ffn_up.weight
    F16
    [1536 6144]
  • blk.20.ffn_norm.bias
    F32
    [1536]
  • blk.20.ffn_norm.weight
    F32
    [1536]
  • blk.20.attn_output.bias
    F32
    [1536]
  • blk.20.attn_output.weight
    F16
    [1536 1536]
  • blk.20.attn_qkv.bias
    F32
    [4608]
  • blk.20.attn_qkv.weight
    F16
    [1536 4608]
  • blk.21
  • blk.21.attn_norm.bias
    F32
    [1536]
  • blk.21.attn_norm.weight
    F32
    [1536]
  • blk.21.ffn_down.bias
    F32
    [1536]
  • blk.21.ffn_down.weight
    F16
    [6144 1536]
  • blk.21.ffn_up.bias
    F32
    [6144]
  • blk.21.ffn_up.weight
    F16
    [1536 6144]
  • blk.21.ffn_norm.bias
    F32
    [1536]
  • blk.21.ffn_norm.weight
    F32
    [1536]
  • blk.21.attn_output.bias
    F32
    [1536]
  • blk.21.attn_output.weight
    F16
    [1536 1536]
  • blk.21.attn_qkv.bias
    F32
    [4608]
  • blk.21.attn_qkv.weight
    F16
    [1536 4608]
  • blk.22
  • blk.22.attn_norm.bias
    F32
    [1536]
  • blk.22.attn_norm.weight
    F32
    [1536]
  • blk.22.ffn_down.bias
    F32
    [1536]
  • blk.22.ffn_down.weight
    F16
    [6144 1536]
  • blk.22.ffn_up.bias
    F32
    [6144]
  • blk.22.ffn_up.weight
    F16
    [1536 6144]
  • blk.22.ffn_norm.bias
    F32
    [1536]
  • blk.22.ffn_norm.weight
    F32
    [1536]
  • blk.22.attn_output.bias
    F32
    [1536]
  • blk.22.attn_output.weight
    F16
    [1536 1536]
  • blk.22.attn_qkv.bias
    F32
    [4608]
  • blk.22.attn_qkv.weight
    F16
    [1536 4608]
  • blk.23
  • blk.23.attn_norm.bias
    F32
    [1536]
  • blk.23.attn_norm.weight
    F32
    [1536]
  • blk.23.ffn_down.bias
    F32
    [1536]
  • blk.23.ffn_down.weight
    F16
    [6144 1536]
  • blk.23.ffn_up.bias
    F32
    [6144]
  • blk.23.ffn_up.weight
    F16
    [1536 6144]
  • blk.23.ffn_norm.bias
    F32
    [1536]
  • blk.23.ffn_norm.weight
    F32
    [1536]
  • blk.23.attn_output.bias
    F32
    [1536]
  • blk.23.attn_output.weight
    F16
    [1536 1536]
  • blk.23.attn_qkv.bias
    F32
    [4608]
  • blk.23.attn_qkv.weight
    F16
    [1536 4608]
  • output_norm.bias
    F32
    [1536]
  • output_norm.weight
    F32
    [1536]
  • output.weight
    F16
    [1536 250880]