embedding
132 Pulls Updated 10 months ago
b33141451456 · 274MB
-
general.architecturenomic-bert
-
general.file_type1
-
nomic-bert.attention.causalfalse
-
nomic-bert.attention.head_count12
-
nomic-bert.attention.layer_norm_epsilon1e-12
-
nomic-bert.block_count12
-
nomic-bert.context_length2048
-
nomic-bert.embedding_length768
-
nomic-bert.feed_forward_length3072
-
nomic-bert.pooling_type1
-
nomic-bert.rope.freq_base1000
-
tokenizer.ggml.bos_token_id101
-
tokenizer.ggml.cls_token_id101
-
tokenizer.ggml.eos_token_id102
-
tokenizer.ggml.mask_token_id103
-
tokenizer.ggml.modelbert
-
tokenizer.ggml.padding_token_id0
-
tokenizer.ggml.scores[-1000, -1000, -1000, -1000, -1000, ...]
-
tokenizer.ggml.seperator_token_id102
-
tokenizer.ggml.token_type[3, 1, 1, 1, 1, ...]
-
tokenizer.ggml.token_type_count2
-
tokenizer.ggml.tokens[[PAD], [unused0], [unused1], [unused2], [unused3], ...]
-
tokenizer.ggml.unknown_token_id100
-
NameTypeShape
-
token_embd.weightF16[768, 30522]
-
blk.0.attn_output.weightF16[768, 768]
-
blk.0.attn_output_norm.biasF32[768]
-
blk.0.attn_output_norm.weightF32[768]
-
blk.0.attn_qkv.weightF16[768, 2304]
-
blk.0.ffn_down.weightF16[3072, 768]
-
blk.0.ffn_gate.weightF16[768, 3072]
-
blk.0.ffn_up.weightF16[768, 3072]
-
blk.0.layer_output_norm.biasF32[768]
-
blk.0.layer_output_norm.weightF32[768]
-
blk.1.attn_output.weightF16[768, 768]
-
blk.1.attn_output_norm.biasF32[768]
-
blk.1.attn_output_norm.weightF32[768]
-
blk.1.attn_qkv.weightF16[768, 2304]
-
blk.1.ffn_down.weightF16[3072, 768]
-
blk.1.ffn_gate.weightF16[768, 3072]
-
blk.1.ffn_up.weightF16[768, 3072]
-
blk.1.layer_output_norm.biasF32[768]
-
blk.1.layer_output_norm.weightF32[768]
-
blk.2.attn_output.weightF16[768, 768]
-
blk.2.attn_output_norm.biasF32[768]
-
blk.2.attn_output_norm.weightF32[768]
-
blk.2.attn_qkv.weightF16[768, 2304]
-
blk.2.ffn_down.weightF16[3072, 768]
-
blk.2.ffn_gate.weightF16[768, 3072]
-
blk.2.ffn_up.weightF16[768, 3072]
-
blk.2.layer_output_norm.biasF32[768]
-
blk.2.layer_output_norm.weightF32[768]
-
blk.3.attn_output.weightF16[768, 768]
-
blk.3.attn_output_norm.biasF32[768]
-
blk.3.attn_output_norm.weightF32[768]
-
blk.3.attn_qkv.weightF16[768, 2304]
-
blk.3.ffn_down.weightF16[3072, 768]
-
blk.3.ffn_gate.weightF16[768, 3072]
-
blk.3.ffn_up.weightF16[768, 3072]
-
blk.3.layer_output_norm.biasF32[768]
-
blk.3.layer_output_norm.weightF32[768]
-
blk.4.attn_output.weightF16[768, 768]
-
blk.4.attn_output_norm.biasF32[768]
-
blk.4.attn_output_norm.weightF32[768]
-
blk.4.attn_qkv.weightF16[768, 2304]
-
blk.4.ffn_down.weightF16[3072, 768]
-
blk.4.ffn_gate.weightF16[768, 3072]
-
blk.4.ffn_up.weightF16[768, 3072]
-
blk.4.layer_output_norm.biasF32[768]
-
blk.4.layer_output_norm.weightF32[768]
-
blk.5.attn_output.weightF16[768, 768]
-
blk.5.attn_output_norm.biasF32[768]
-
blk.5.attn_output_norm.weightF32[768]
-
blk.5.attn_qkv.weightF16[768, 2304]
-
blk.5.ffn_down.weightF16[3072, 768]
-
blk.5.ffn_gate.weightF16[768, 3072]
-
blk.5.ffn_up.weightF16[768, 3072]
-
blk.5.layer_output_norm.biasF32[768]
-
blk.5.layer_output_norm.weightF32[768]
-
blk.6.attn_output.weightF16[768, 768]
-
blk.6.attn_output_norm.biasF32[768]
-
blk.6.attn_output_norm.weightF32[768]
-
blk.6.attn_qkv.weightF16[768, 2304]
-
blk.6.ffn_down.weightF16[3072, 768]
-
blk.6.ffn_gate.weightF16[768, 3072]
-
blk.6.ffn_up.weightF16[768, 3072]
-
blk.6.layer_output_norm.biasF32[768]
-
blk.6.layer_output_norm.weightF32[768]
-
blk.7.attn_output.weightF16[768, 768]
-
blk.7.attn_output_norm.biasF32[768]
-
blk.7.attn_output_norm.weightF32[768]
-
blk.7.attn_qkv.weightF16[768, 2304]
-
blk.7.ffn_down.weightF16[3072, 768]
-
blk.7.ffn_gate.weightF16[768, 3072]
-
blk.7.ffn_up.weightF16[768, 3072]
-
blk.7.layer_output_norm.biasF32[768]
-
blk.7.layer_output_norm.weightF32[768]
-
blk.8.attn_output.weightF16[768, 768]
-
blk.8.attn_output_norm.biasF32[768]
-
blk.8.attn_output_norm.weightF32[768]
-
blk.8.attn_qkv.weightF16[768, 2304]
-
blk.8.ffn_down.weightF16[3072, 768]
-
blk.8.ffn_gate.weightF16[768, 3072]
-
blk.8.ffn_up.weightF16[768, 3072]
-
blk.8.layer_output_norm.biasF32[768]
-
blk.8.layer_output_norm.weightF32[768]
-
blk.9.attn_output.weightF16[768, 768]
-
blk.9.attn_output_norm.biasF32[768]
-
blk.9.attn_output_norm.weightF32[768]
-
blk.9.attn_qkv.weightF16[768, 2304]
-
blk.9.ffn_down.weightF16[3072, 768]
-
blk.9.ffn_gate.weightF16[768, 3072]
-
blk.9.ffn_up.weightF16[768, 3072]
-
blk.9.layer_output_norm.biasF32[768]
-
blk.9.layer_output_norm.weightF32[768]
-
blk.10.attn_output.weightF16[768, 768]
-
blk.10.attn_output_norm.biasF32[768]
-
blk.10.attn_output_norm.weightF32[768]
-
blk.10.attn_qkv.weightF16[768, 2304]
-
blk.10.ffn_down.weightF16[3072, 768]
-
blk.10.ffn_gate.weightF16[768, 3072]
-
blk.10.ffn_up.weightF16[768, 3072]
-
blk.10.layer_output_norm.biasF32[768]
-
blk.10.layer_output_norm.weightF32[768]
-
blk.11.attn_output.weightF16[768, 768]
-
blk.11.attn_output_norm.biasF32[768]
-
blk.11.attn_output_norm.weightF32[768]
-
blk.11.attn_qkv.weightF16[768, 2304]
-
blk.11.ffn_down.weightF16[3072, 768]
-
blk.11.ffn_gate.weightF16[768, 3072]
-
blk.11.ffn_up.weightF16[768, 3072]
-
blk.11.layer_output_norm.biasF32[768]
-
blk.11.layer_output_norm.weightF32[768]
-
token_embd_norm.biasF32[768]
-
token_embd_norm.weightF32[768]
-
token_types.weightF32[768, 2]
Metadata
Tensor
blk.0
blk.1
blk.2
blk.3
blk.4
blk.5
blk.6
blk.7
blk.8
blk.9
blk.10
blk.11