sppandita85/ llm-from-scratch:latest

3 1 week ago

... /
model
c98aa6fe57be · 56MB
    Metadata
  • general.architecture
    gpt2
  • general.file_type
    F32
  • gpt2.attention.head_count
    4
  • gpt2.attention.layer_norm_epsilon
    1e-05
  • gpt2.block_count
    4
  • gpt2.context_length
    128
  • gpt2.embedding_length
    128
  • gpt2.feed_forward_length
    512
  • tokenizer.ggml.add_bos_token
    false
  • tokenizer.ggml.add_eos_token
    false
  • tokenizer.ggml.bos_token_id
    50256
  • tokenizer.ggml.eos_token_id
    50256
  • tokenizer.ggml.merges
    [Ġ t, Ġ a, h e, i n, r e, ...]
  • tokenizer.ggml.model
    gpt2
  • tokenizer.ggml.pre
    gpt-2
  • tokenizer.ggml.token_type
    [1, 1, 1, 1, 1, ...]
  • tokenizer.ggml.tokens
    [!, ", #, $, %, ...]
  • tokenizer.ggml.unknown_token_id
    50256
    • Tensor
  • token_embd.weight
    F32
    [128, 50257]
    • blk.0
  • blk.0.attn_norm.bias
    F32
    [128]
  • blk.0.attn_norm.weight
    F32
    [128]
  • blk.0.attn_output.bias
    F32
    [128]
  • blk.0.attn_output.weight
    F32
    [128, 128]
  • blk.0.attn_qkv.bias
    F32
    [384]
  • blk.0.attn_qkv.weight
    F32
    [128, 384]
  • blk.0.ffn_down.bias
    F32
    [128]
  • blk.0.ffn_down.weight
    F32
    [512, 128]
  • blk.0.ffn_norm.bias
    F32
    [128]
  • blk.0.ffn_norm.weight
    F32
    [128]
  • blk.0.ffn_up.bias
    F32
    [512]
  • blk.0.ffn_up.weight
    F32
    [128, 512]
    • blk.1
  • blk.1.attn_norm.bias
    F32
    [128]
  • blk.1.attn_norm.weight
    F32
    [128]
  • blk.1.attn_output.bias
    F32
    [128]
  • blk.1.attn_output.weight
    F32
    [128, 128]
  • blk.1.attn_qkv.bias
    F32
    [384]
  • blk.1.attn_qkv.weight
    F32
    [128, 384]
  • blk.1.ffn_down.bias
    F32
    [128]
  • blk.1.ffn_down.weight
    F32
    [512, 128]
  • blk.1.ffn_norm.bias
    F32
    [128]
  • blk.1.ffn_norm.weight
    F32
    [128]
  • blk.1.ffn_up.bias
    F32
    [512]
  • blk.1.ffn_up.weight
    F32
    [128, 512]
    • blk.2
  • blk.2.attn_norm.bias
    F32
    [128]
  • blk.2.attn_norm.weight
    F32
    [128]
  • blk.2.attn_output.bias
    F32
    [128]
  • blk.2.attn_output.weight
    F32
    [128, 128]
  • blk.2.attn_qkv.bias
    F32
    [384]
  • blk.2.attn_qkv.weight
    F32
    [128, 384]
  • blk.2.ffn_down.bias
    F32
    [128]
  • blk.2.ffn_down.weight
    F32
    [512, 128]
  • blk.2.ffn_norm.bias
    F32
    [128]
  • blk.2.ffn_norm.weight
    F32
    [128]
  • blk.2.ffn_up.bias
    F32
    [512]
  • blk.2.ffn_up.weight
    F32
    [128, 512]
    • blk.3
  • blk.3.attn_norm.bias
    F32
    [128]
  • blk.3.attn_norm.weight
    F32
    [128]
  • blk.3.attn_output.bias
    F32
    [128]
  • blk.3.attn_output.weight
    F32
    [128, 128]
  • blk.3.attn_qkv.bias
    F32
    [384]
  • blk.3.attn_qkv.weight
    F32
    [128, 384]
  • blk.3.ffn_down.bias
    F32
    [128]
  • blk.3.ffn_down.weight
    F32
    [512, 128]
  • blk.3.ffn_norm.bias
    F32
    [128]
  • blk.3.ffn_norm.weight
    F32
    [128]
  • blk.3.ffn_up.bias
    F32
    [512]
  • blk.3.ffn_up.weight
    F32
    [128, 512]
  • output.weight
    F32
    [128, 50257]
  • output_norm.bias
    F32
    [128]
  • position_embd.weight
    F32
    [128, 128]
  • output_norm.weight
    F32
    [128]