A set of EleutherAI models trained on the Pile.
4 Pulls Updated 5 days ago
d7824bd3f55f · 17MB
-
general.architecturegptneox
-
general.file_typeQ8_0
-
gptneox.attention.head_count4
-
gptneox.attention.layer_norm_epsilon1e-05
-
gptneox.block_count6
-
gptneox.context_length2048
-
gptneox.embedding_length128
-
gptneox.feed_forward_length512
-
gptneox.rope.dimension_count8
-
gptneox.use_parallel_residualtrue
-
tokenizer.ggml.bos_token_id0
-
tokenizer.ggml.eos_token_id0
-
tokenizer.ggml.merges[Ġ Ġ Ġ t Ġ a h e i n ...]
-
tokenizer.ggml.modelgpt2
-
tokenizer.ggml.preolmo
-
tokenizer.ggml.token_type[3 3 1 1 1 ...]
-
tokenizer.ggml.tokens[<|endoftext|> <|padding|> ! " # ...]
-
tokenizer.ggml.unknown_token_id0
-
NameTypeShape
-
token_embd.weightQ8_0[128 50304]
-
blk.0.attn_output.biasF32[128]
-
blk.0.attn_output.weightQ8_0[128 128]
-
blk.0.attn_qkv.biasF32[384]
-
blk.0.attn_qkv.weightQ8_0[128 384]
-
blk.0.attn_norm.biasF32[128]
-
blk.0.attn_norm.weightF32[128]
-
blk.0.ffn_down.biasF32[128]
-
blk.0.ffn_down.weightQ8_0[512 128]
-
blk.0.ffn_up.biasF32[512]
-
blk.0.ffn_up.weightQ8_0[128 512]
-
blk.0.ffn_norm.biasF32[128]
-
blk.0.ffn_norm.weightF32[128]
-
blk.1.attn_output.biasF32[128]
-
blk.1.attn_output.weightQ8_0[128 128]
-
blk.1.attn_qkv.biasF32[384]
-
blk.1.attn_qkv.weightQ8_0[128 384]
-
blk.1.attn_norm.biasF32[128]
-
blk.1.attn_norm.weightF32[128]
-
blk.1.ffn_down.biasF32[128]
-
blk.1.ffn_down.weightQ8_0[512 128]
-
blk.1.ffn_up.biasF32[512]
-
blk.1.ffn_up.weightQ8_0[128 512]
-
blk.1.ffn_norm.biasF32[128]
-
blk.1.ffn_norm.weightF32[128]
-
blk.2.attn_output.biasF32[128]
-
blk.2.attn_output.weightQ8_0[128 128]
-
blk.2.attn_qkv.biasF32[384]
-
blk.2.attn_qkv.weightQ8_0[128 384]
-
blk.2.attn_norm.biasF32[128]
-
blk.2.attn_norm.weightF32[128]
-
blk.2.ffn_down.biasF32[128]
-
blk.2.ffn_down.weightQ8_0[512 128]
-
blk.2.ffn_up.biasF32[512]
-
blk.2.ffn_up.weightQ8_0[128 512]
-
blk.2.ffn_norm.biasF32[128]
-
blk.2.ffn_norm.weightF32[128]
-
blk.3.attn_output.biasF32[128]
-
blk.3.attn_output.weightQ8_0[128 128]
-
blk.3.attn_qkv.biasF32[384]
-
blk.3.attn_qkv.weightQ8_0[128 384]
-
blk.3.attn_norm.biasF32[128]
-
blk.3.attn_norm.weightF32[128]
-
blk.3.ffn_down.biasF32[128]
-
blk.3.ffn_down.weightQ8_0[512 128]
-
blk.3.ffn_up.biasF32[512]
-
blk.3.ffn_up.weightQ8_0[128 512]
-
blk.3.ffn_norm.biasF32[128]
-
blk.3.ffn_norm.weightF32[128]
-
blk.4.attn_output.biasF32[128]
-
blk.4.attn_output.weightQ8_0[128 128]
-
blk.4.attn_qkv.biasF32[384]
-
blk.4.attn_qkv.weightQ8_0[128 384]
-
blk.4.attn_norm.biasF32[128]
-
blk.4.attn_norm.weightF32[128]
-
blk.4.ffn_down.biasF32[128]
-
blk.4.ffn_down.weightQ8_0[512 128]
-
blk.4.ffn_up.biasF32[512]
-
blk.4.ffn_up.weightQ8_0[128 512]
-
blk.4.ffn_norm.biasF32[128]
-
blk.4.ffn_norm.weightF32[128]
-
blk.5.attn_output.biasF32[128]
-
blk.5.attn_output.weightQ8_0[128 128]
-
blk.5.attn_qkv.biasF32[384]
-
blk.5.attn_qkv.weightQ8_0[128 384]
-
blk.5.attn_norm.biasF32[128]
-
blk.5.attn_norm.weightF32[128]
-
blk.5.ffn_down.biasF32[128]
-
blk.5.ffn_down.weightQ8_0[512 128]
-
blk.5.ffn_up.biasF32[512]
-
blk.5.ffn_up.weightQ8_0[128 512]
-
blk.5.ffn_norm.biasF32[128]
-
blk.5.ffn_norm.weightF32[128]
-
output.weightQ8_0[128 50304]
-
output_norm.biasF32[128]
-
output_norm.weightF32[128]
Metadata
Tensors
blk.0
blk.1
blk.2
blk.3
blk.4
blk.5