deploy1001 / fine_tune

1 Pull Updated 2 months ago

... /
model
6a40c815a788 · 1.6MB
    Metadata
  • general.architecture
    llama
  • general.file_type
    F16
  • llama.attention.head_count
    32
  • llama.attention.head_count_kv
    32
  • llama.attention.layer_norm_rms_epsilon
    1e-05
  • llama.block_count
    32
  • llama.context_length
    4096
  • llama.embedding_length
    4096
  • llama.feed_forward_length
    11008
  • llama.rope.dimension_count
    128
  • llama.rope.freq_base
    0
  • llama.vocab_size
    32001
  • tokenizer.ggml.bos_token_id
    1
  • tokenizer.ggml.eos_token_id
    2
  • tokenizer.ggml.merges
    [▁ t e r i n ▁ a e n ...]
  • tokenizer.ggml.model
    gpt2
  • tokenizer.ggml.pre
    default
  • tokenizer.ggml.token_type
    [3 3 3 1 1 ...]
  • tokenizer.ggml.tokens
    [<unk> <s> </s> <0x00> <0x01> ...]
  • tokenizer.ggml.unknown_token_id
    0