clore/
grok-1:latest

41 2 months ago

a1e2f589771b · 130GB
    Metadata
  • general.architecture
    grok
  • general.file_type
    IQ3_XS
  • grok.attention.head_count
    48
  • grok.attention.head_count_kv
    8
  • grok.attention.layer_norm_rms_epsilon
    1e-05
  • grok.block_count
    64
  • grok.context_length
    8192
  • grok.embedding_length
    6144
  • grok.expert_count
    8
  • grok.expert_used_count
    2
  • grok.feed_forward_length
    32768
  • grok.rope.freq_base
    10000
  • tokenizer.ggml.model
    llama
  • tokenizer.ggml.scores
    [0, 0, 0, 0, 0, ...]
  • tokenizer.ggml.token_type
    [3, 3, 3, 2, 1, ...]
  • tokenizer.ggml.tokens
    [[PAD], [BOS], [EOS], [UNK], , ...]
  • split.count
    9
  • split.no
    0
  • split.tensors.count
    2114
  • Tensor
  • token_embd.weight
    Q6_K
    [6144, 131072]
  • blk.0
  • blk.0.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.0.attn_norm.weight
    F32
    [6144]
  • blk.0.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.0.attn_output_norm.weight
    F32
    [6144]
  • blk.0.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.0.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.0.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.0.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.0.ffn_norm.weight
    F32
    [6144]
  • blk.0.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.0.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.0.layer_output_norm.weight
    F32
    [6144]
  • blk.1
  • blk.1.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.1.attn_norm.weight
    F32
    [6144]
  • blk.1.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.1.attn_output_norm.weight
    F32
    [6144]
  • blk.1.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.1.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.1.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.1.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.1.ffn_norm.weight
    F32
    [6144]
  • blk.1.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.1.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.1.layer_output_norm.weight
    F32
    [6144]
  • blk.2
  • blk.2.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.2.attn_norm.weight
    F32
    [6144]
  • blk.2.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.2.attn_output_norm.weight
    F32
    [6144]
  • blk.2.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.2.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.2.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.2.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.2.ffn_norm.weight
    F32
    [6144]
  • blk.2.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.2.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.2.layer_output_norm.weight
    F32
    [6144]
  • blk.3
  • blk.3.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.3.attn_norm.weight
    F32
    [6144]
  • blk.3.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.3.attn_output_norm.weight
    F32
    [6144]
  • blk.3.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.3.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.3.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.3.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.3.ffn_norm.weight
    F32
    [6144]
  • blk.3.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.3.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.3.layer_output_norm.weight
    F32
    [6144]
  • blk.4
  • blk.4.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.4.attn_norm.weight
    F32
    [6144]
  • blk.4.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.4.attn_output_norm.weight
    F32
    [6144]
  • blk.4.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.4.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.4.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.4.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.4.ffn_norm.weight
    F32
    [6144]
  • blk.4.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.4.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.4.layer_output_norm.weight
    F32
    [6144]
  • blk.5
  • blk.5.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.5.attn_norm.weight
    F32
    [6144]
  • blk.5.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.5.attn_output_norm.weight
    F32
    [6144]
  • blk.5.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.5.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.5.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.5.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.5.ffn_norm.weight
    F32
    [6144]
  • blk.5.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.5.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.5.layer_output_norm.weight
    F32
    [6144]
  • blk.6
  • blk.6.attn_k.weight
    Q8_0
    [6144, 1024]
  • blk.6.attn_norm.weight
    F32
    [6144]
  • blk.6.attn_output.weight
    Q5_K
    [6144, 6144]
  • blk.6.attn_output_norm.weight
    F32
    [6144]
  • blk.6.attn_q.weight
    IQ3_XXS
    [6144, 6144]
  • blk.6.attn_v.weight
    Q8_0
    [6144, 1024]
  • blk.6.ffn_down.0.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.1.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.2.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.3.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.4.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.5.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.6.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_down.7.weight
    IQ3_S
    [32768, 6144]
  • blk.6.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate.7.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_gate_inp.weight
    F32
    [6144, 8]
  • blk.6.ffn_norm.weight
    F32
    [6144]
  • blk.6.ffn_up.0.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.1.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.2.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.3.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.4.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.5.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.6.weight
    IQ3_S
    [6144, 32768]
  • blk.6.ffn_up.7.weight
    IQ3_S
    [6144, 32768]
  • blk.6.layer_output_norm.weight
    F32
    [6144]
  • blk.7
  • blk.7.ffn_gate.0.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.1.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.2.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.3.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.4.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.5.weight
    IQ3_S
    [6144, 32768]
  • blk.7.ffn_gate.6.weight
    IQ3_S
    [6144, 32768]
  • output_norm.weight
    F32
    [6144]