TCYZ/ cokertme2:5m

45 1 month ago

Çökertme 2, Türkçe doğal dil işleme görevleri için optimize edilmiş, ultra-nano ölçekli bir dil modeli ailesidir.

0.8m 1.6m 5m 6m 14m 100m
... /
model
718a77891d67 · 21MB
    Metadata
  • general.architecture
    llama
  • llama.attention.head_count
    1
  • llama.attention.head_count_kv
    1
  • llama.attention.layer_norm_rms_epsilon
    1e-05
  • llama.block_count
    1
  • llama.context_length
    256
  • llama.embedding_length
    50
  • llama.feed_forward_length
    133
  • tokenizer.ggml.model
    llama
  • tokenizer.ggml.scores
    [0, 0, 0, 0, 0, ...]
  • tokenizer.ggml.token_type
    [1, 1, 1, 1, 1, ...]
  • tokenizer.ggml.tokens
    [!, ", #, $, %, ...]
    • Tensor
  • token_embd.weight
    F32
    [50, 50257]
    • blk.0
  • blk.0.attn_k.weight
    F32
    [50, 50]
  • blk.0.attn_norm.weight
    F32
    [50]
  • blk.0.attn_output.weight
    F32
    [50, 50]
  • blk.0.attn_q.weight
    F32
    [50, 50]
  • blk.0.attn_v.weight
    F32
    [50, 50]
  • blk.0.ffn_down.weight
    F32
    [133, 50]
  • blk.0.ffn_gate.weight
    F32
    [50, 133]
  • blk.0.ffn_norm.weight
    F32
    [50]
  • blk.0.ffn_up.weight
    F32
    [50, 133]
  • output.weight
    F32
    [50, 50257]
  • output_norm.weight
    F32
    [50]