This model is a fine-tuned version of allenai/OLMo-1B-hf on the HuggingFaceH4/ultrachat_200k dataset
28 Pulls Updated 5 months ago
5834074e37a0 · 847MB
-
general.architectureolmo
-
general.file_type17
-
general.namemodel
-
general.quantization_version2
-
olmo.attention.head_count16
-
olmo.attention.head_count_kv16
-
olmo.attention.layer_norm_epsilon1e-05
-
olmo.block_count16
-
olmo.context_length2048
-
olmo.embedding_length2048
-
olmo.feed_forward_length8192
-
olmo.rope.freq_base10000
-
tokenizer.ggml.add_bos_tokenfalse
-
tokenizer.ggml.add_eos_tokenfalse
-
tokenizer.ggml.eos_token_id50279
-
tokenizer.ggml.merges[Ġ Ġ, Ġ t, Ġ a, h e, i n, ...]
-
tokenizer.ggml.modelgpt2
-
tokenizer.ggml.padding_token_id1
-
tokenizer.ggml.preolmo
-
tokenizer.ggml.token_type[4, 3, 1, 1, 1, ...]
-
tokenizer.ggml.tokens[|||IP_ADDRESS|||, <|padding|>, !, ", #, ...]
-
NameTypeShape
-
token_embd.weightQ6_K[2048, 50304]
-
blk.0.attn_k.weightQ5_K[2048, 2048]
-
blk.0.attn_output.weightQ5_K[2048, 2048]
-
blk.0.attn_q.weightQ5_K[2048, 2048]
-
blk.0.attn_v.weightQ6_K[2048, 2048]
-
blk.0.ffn_down.weightQ6_K[8192, 2048]
-
blk.0.ffn_gate.weightQ5_K[2048, 8192]
-
blk.0.ffn_up.weightQ5_K[2048, 8192]
-
blk.1.attn_k.weightQ5_K[2048, 2048]
-
blk.1.attn_output.weightQ5_K[2048, 2048]
-
blk.1.attn_q.weightQ5_K[2048, 2048]
-
blk.1.attn_v.weightQ6_K[2048, 2048]
-
blk.1.ffn_down.weightQ6_K[8192, 2048]
-
blk.1.ffn_gate.weightQ5_K[2048, 8192]
-
blk.1.ffn_up.weightQ5_K[2048, 8192]
-
blk.2.attn_k.weightQ5_K[2048, 2048]
-
blk.2.attn_output.weightQ5_K[2048, 2048]
-
blk.2.attn_q.weightQ5_K[2048, 2048]
-
blk.2.attn_v.weightQ5_K[2048, 2048]
-
blk.2.ffn_down.weightQ5_K[8192, 2048]
-
blk.2.ffn_gate.weightQ5_K[2048, 8192]
-
blk.2.ffn_up.weightQ5_K[2048, 8192]
-
blk.3.attn_k.weightQ5_K[2048, 2048]
-
blk.3.attn_output.weightQ5_K[2048, 2048]
-
blk.3.attn_q.weightQ5_K[2048, 2048]
-
blk.3.attn_v.weightQ5_K[2048, 2048]
-
blk.3.ffn_down.weightQ5_K[8192, 2048]
-
blk.3.ffn_gate.weightQ5_K[2048, 8192]
-
blk.3.ffn_up.weightQ5_K[2048, 8192]
-
blk.4.attn_k.weightQ5_K[2048, 2048]
-
blk.4.attn_output.weightQ5_K[2048, 2048]
-
blk.4.attn_q.weightQ5_K[2048, 2048]
-
blk.4.attn_v.weightQ6_K[2048, 2048]
-
blk.4.ffn_down.weightQ6_K[8192, 2048]
-
blk.4.ffn_gate.weightQ5_K[2048, 8192]
-
blk.4.ffn_up.weightQ5_K[2048, 8192]
-
blk.5.attn_k.weightQ5_K[2048, 2048]
-
blk.5.attn_output.weightQ5_K[2048, 2048]
-
blk.5.attn_q.weightQ5_K[2048, 2048]
-
blk.5.attn_v.weightQ5_K[2048, 2048]
-
blk.5.ffn_down.weightQ5_K[8192, 2048]
-
blk.5.ffn_gate.weightQ5_K[2048, 8192]
-
blk.5.ffn_up.weightQ5_K[2048, 8192]
-
blk.6.attn_k.weightQ5_K[2048, 2048]
-
blk.6.attn_output.weightQ5_K[2048, 2048]
-
blk.6.attn_q.weightQ5_K[2048, 2048]
-
blk.6.attn_v.weightQ5_K[2048, 2048]
-
blk.6.ffn_down.weightQ5_K[8192, 2048]
-
blk.6.ffn_gate.weightQ5_K[2048, 8192]
-
blk.6.ffn_up.weightQ5_K[2048, 8192]
-
blk.7.attn_k.weightQ5_K[2048, 2048]
-
blk.7.attn_output.weightQ5_K[2048, 2048]
-
blk.7.attn_q.weightQ5_K[2048, 2048]
-
blk.7.attn_v.weightQ6_K[2048, 2048]
-
blk.7.ffn_down.weightQ6_K[8192, 2048]
-
blk.7.ffn_gate.weightQ5_K[2048, 8192]
-
blk.7.ffn_up.weightQ5_K[2048, 8192]
-
blk.8.attn_k.weightQ5_K[2048, 2048]
-
blk.8.attn_output.weightQ5_K[2048, 2048]
-
blk.8.attn_q.weightQ5_K[2048, 2048]
-
blk.8.attn_v.weightQ6_K[2048, 2048]
-
blk.8.ffn_down.weightQ6_K[8192, 2048]
-
blk.8.ffn_gate.weightQ5_K[2048, 8192]
-
blk.8.ffn_up.weightQ5_K[2048, 8192]
-
blk.9.attn_k.weightQ5_K[2048, 2048]
-
blk.9.attn_output.weightQ5_K[2048, 2048]
-
blk.9.attn_q.weightQ5_K[2048, 2048]
-
blk.9.attn_v.weightQ6_K[2048, 2048]
-
blk.9.ffn_down.weightQ6_K[8192, 2048]
-
blk.9.ffn_gate.weightQ5_K[2048, 8192]
-
blk.9.ffn_up.weightQ5_K[2048, 8192]
-
blk.10.attn_k.weightQ5_K[2048, 2048]
-
blk.10.attn_output.weightQ5_K[2048, 2048]
-
blk.10.attn_q.weightQ5_K[2048, 2048]
-
blk.10.attn_v.weightQ5_K[2048, 2048]
-
blk.10.ffn_down.weightQ5_K[8192, 2048]
-
blk.10.ffn_gate.weightQ5_K[2048, 8192]
-
blk.10.ffn_up.weightQ5_K[2048, 8192]
-
blk.11.attn_k.weightQ5_K[2048, 2048]
-
blk.11.attn_output.weightQ5_K[2048, 2048]
-
blk.11.attn_q.weightQ5_K[2048, 2048]
-
blk.11.attn_v.weightQ5_K[2048, 2048]
-
blk.11.ffn_down.weightQ5_K[8192, 2048]
-
blk.11.ffn_gate.weightQ5_K[2048, 8192]
-
blk.11.ffn_up.weightQ5_K[2048, 8192]
-
blk.12.attn_k.weightQ5_K[2048, 2048]
-
blk.12.attn_output.weightQ5_K[2048, 2048]
-
blk.12.attn_q.weightQ5_K[2048, 2048]
-
blk.12.attn_v.weightQ6_K[2048, 2048]
-
blk.12.ffn_down.weightQ6_K[8192, 2048]
-
blk.12.ffn_gate.weightQ5_K[2048, 8192]
-
blk.12.ffn_up.weightQ5_K[2048, 8192]
-
blk.13.attn_k.weightQ5_K[2048, 2048]
-
blk.13.attn_output.weightQ5_K[2048, 2048]
-
blk.13.attn_q.weightQ5_K[2048, 2048]
-
blk.13.attn_v.weightQ5_K[2048, 2048]
-
blk.13.ffn_down.weightQ5_K[8192, 2048]
-
blk.13.ffn_gate.weightQ5_K[2048, 8192]
-
blk.13.ffn_up.weightQ5_K[2048, 8192]
-
blk.14.attn_k.weightQ5_K[2048, 2048]
-
blk.14.attn_output.weightQ5_K[2048, 2048]
-
blk.14.attn_q.weightQ5_K[2048, 2048]
-
blk.14.attn_v.weightQ5_K[2048, 2048]
-
blk.14.ffn_down.weightQ5_K[8192, 2048]
-
blk.14.ffn_gate.weightQ5_K[2048, 8192]
-
blk.14.ffn_up.weightQ5_K[2048, 8192]
-
blk.15.attn_k.weightQ5_K[2048, 2048]
-
blk.15.attn_output.weightQ5_K[2048, 2048]
-
blk.15.attn_q.weightQ5_K[2048, 2048]
-
blk.15.attn_v.weightQ6_K[2048, 2048]
-
blk.15.ffn_down.weightQ6_K[8192, 2048]
-
blk.15.ffn_gate.weightQ5_K[2048, 8192]
-
blk.15.ffn_up.weightQ5_K[2048, 8192]
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
blk.12
blk.13
blk.14
blk.15