Qwen 1.5 is a series of large language models by Alibaba Cloud spanning from 0.5B to 110B parameters
0.5b
1.8b
4b
7b
14b
32b
72b
110b
4.1M Pulls Updated 7 months ago
9a9f0c3f0202 · 1.4GB
-
general.architectureqwen
-
general.file_type17
-
general.nameQwen
-
general.quantization_version2
-
qwen.attention.head_count16
-
qwen.attention.layer_norm_rms_epsilon1e-06
-
qwen.block_count24
-
qwen.context_length8192
-
qwen.embedding_length2048
-
qwen.feed_forward_length11008
-
qwen.rope.dimension_count128
-
qwen.rope.freq_base10000
-
tokenizer.ggml.bos_token_id151643
-
tokenizer.ggml.eos_token_id151643
-
tokenizer.ggml.merges[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
-
tokenizer.ggml.modelgpt2
-
tokenizer.ggml.token_type[1, 1, 1, 1, 1, ...]
-
tokenizer.ggml.tokens[!, ", #, $, %, ...]
-
tokenizer.ggml.unknown_token_id151643
-
NameTypeShape
-
token_embd.weightQ5_K[2048, 151936]
-
blk.0.attn_norm.weightF32[2048]
-
blk.0.attn_output.weightQ5_K[2048, 2048]
-
blk.0.attn_qkv.biasF32[6144]
-
blk.0.attn_qkv.weightQ6_K[2048, 6144]
-
blk.0.ffn_down.weightQ8_0[5504, 2048]
-
blk.0.ffn_gate.weightQ5_K[2048, 5504]
-
blk.0.ffn_norm.weightF32[2048]
-
blk.0.ffn_up.weightQ5_K[2048, 5504]
-
blk.1.attn_norm.weightF32[2048]
-
blk.1.attn_output.weightQ5_K[2048, 2048]
-
blk.1.attn_qkv.biasF32[6144]
-
blk.1.attn_qkv.weightQ6_K[2048, 6144]
-
blk.1.ffn_down.weightQ8_0[5504, 2048]
-
blk.1.ffn_gate.weightQ5_K[2048, 5504]
-
blk.1.ffn_norm.weightF32[2048]
-
blk.1.ffn_up.weightQ5_K[2048, 5504]
-
blk.2.attn_norm.weightF32[2048]
-
blk.2.attn_output.weightQ5_K[2048, 2048]
-
blk.2.attn_qkv.biasF32[6144]
-
blk.2.attn_qkv.weightQ6_K[2048, 6144]
-
blk.2.ffn_down.weightQ5_1[5504, 2048]
-
blk.2.ffn_gate.weightQ5_K[2048, 5504]
-
blk.2.ffn_norm.weightF32[2048]
-
blk.2.ffn_up.weightQ5_K[2048, 5504]
-
blk.3.attn_norm.weightF32[2048]
-
blk.3.attn_output.weightQ5_K[2048, 2048]
-
blk.3.attn_qkv.biasF32[6144]
-
blk.3.attn_qkv.weightQ6_K[2048, 6144]
-
blk.3.ffn_down.weightQ5_1[5504, 2048]
-
blk.3.ffn_gate.weightQ5_K[2048, 5504]
-
blk.3.ffn_norm.weightF32[2048]
-
blk.3.ffn_up.weightQ5_K[2048, 5504]
-
blk.4.attn_norm.weightF32[2048]
-
blk.4.attn_output.weightQ5_K[2048, 2048]
-
blk.4.attn_qkv.biasF32[6144]
-
blk.4.attn_qkv.weightQ6_K[2048, 6144]
-
blk.4.ffn_down.weightQ8_0[5504, 2048]
-
blk.4.ffn_gate.weightQ5_K[2048, 5504]
-
blk.4.ffn_norm.weightF32[2048]
-
blk.4.ffn_up.weightQ5_K[2048, 5504]
-
blk.5.attn_norm.weightF32[2048]
-
blk.5.attn_output.weightQ5_K[2048, 2048]
-
blk.5.attn_qkv.biasF32[6144]
-
blk.5.attn_qkv.weightQ6_K[2048, 6144]
-
blk.5.ffn_down.weightQ5_1[5504, 2048]
-
blk.5.ffn_gate.weightQ5_K[2048, 5504]
-
blk.5.ffn_norm.weightF32[2048]
-
blk.5.ffn_up.weightQ5_K[2048, 5504]
-
blk.6.attn_norm.weightF32[2048]
-
blk.6.attn_output.weightQ5_K[2048, 2048]
-
blk.6.attn_qkv.biasF32[6144]
-
blk.6.attn_qkv.weightQ6_K[2048, 6144]
-
blk.6.ffn_down.weightQ5_1[5504, 2048]
-
blk.6.ffn_gate.weightQ5_K[2048, 5504]
-
blk.6.ffn_norm.weightF32[2048]
-
blk.6.ffn_up.weightQ5_K[2048, 5504]
-
blk.7.attn_norm.weightF32[2048]
-
blk.7.attn_output.weightQ5_K[2048, 2048]
-
blk.7.attn_qkv.biasF32[6144]
-
blk.7.attn_qkv.weightQ6_K[2048, 6144]
-
blk.7.ffn_down.weightQ8_0[5504, 2048]
-
blk.7.ffn_gate.weightQ5_K[2048, 5504]
-
blk.7.ffn_norm.weightF32[2048]
-
blk.7.ffn_up.weightQ5_K[2048, 5504]
-
blk.8.attn_norm.weightF32[2048]
-
blk.8.attn_output.weightQ5_K[2048, 2048]
-
blk.8.attn_qkv.biasF32[6144]
-
blk.8.attn_qkv.weightQ6_K[2048, 6144]
-
blk.8.ffn_down.weightQ5_1[5504, 2048]
-
blk.8.ffn_gate.weightQ5_K[2048, 5504]
-
blk.8.ffn_norm.weightF32[2048]
-
blk.8.ffn_up.weightQ5_K[2048, 5504]
-
blk.9.attn_norm.weightF32[2048]
-
blk.9.attn_output.weightQ5_K[2048, 2048]
-
blk.9.attn_qkv.biasF32[6144]
-
blk.9.attn_qkv.weightQ6_K[2048, 6144]
-
blk.9.ffn_down.weightQ5_1[5504, 2048]
-
blk.9.ffn_gate.weightQ5_K[2048, 5504]
-
blk.9.ffn_norm.weightF32[2048]
-
blk.9.ffn_up.weightQ5_K[2048, 5504]
-
blk.10.attn_norm.weightF32[2048]
-
blk.10.attn_output.weightQ5_K[2048, 2048]
-
blk.10.attn_qkv.biasF32[6144]
-
blk.10.attn_qkv.weightQ6_K[2048, 6144]
-
blk.10.ffn_down.weightQ8_0[5504, 2048]
-
blk.10.ffn_gate.weightQ5_K[2048, 5504]
-
blk.10.ffn_norm.weightF32[2048]
-
blk.10.ffn_up.weightQ5_K[2048, 5504]
-
blk.11.attn_norm.weightF32[2048]
-
blk.11.attn_output.weightQ5_K[2048, 2048]
-
blk.11.attn_qkv.biasF32[6144]
-
blk.11.attn_qkv.weightQ6_K[2048, 6144]
-
blk.11.ffn_down.weightQ5_1[5504, 2048]
-
blk.11.ffn_gate.weightQ5_K[2048, 5504]
-
blk.11.ffn_norm.weightF32[2048]
-
blk.11.ffn_up.weightQ5_K[2048, 5504]
-
blk.12.attn_norm.weightF32[2048]
-
blk.12.attn_output.weightQ5_K[2048, 2048]
-
blk.12.attn_qkv.biasF32[6144]
-
blk.12.attn_qkv.weightQ6_K[2048, 6144]
-
blk.12.ffn_down.weightQ5_1[5504, 2048]
-
blk.12.ffn_gate.weightQ5_K[2048, 5504]
-
blk.12.ffn_norm.weightF32[2048]
-
blk.12.ffn_up.weightQ5_K[2048, 5504]
-
blk.13.attn_norm.weightF32[2048]
-
blk.13.attn_output.weightQ5_K[2048, 2048]
-
blk.13.attn_qkv.biasF32[6144]
-
blk.13.attn_qkv.weightQ6_K[2048, 6144]
-
blk.13.ffn_down.weightQ8_0[5504, 2048]
-
blk.13.ffn_gate.weightQ5_K[2048, 5504]
-
blk.13.ffn_norm.weightF32[2048]
-
blk.13.ffn_up.weightQ5_K[2048, 5504]
-
blk.14.attn_norm.weightF32[2048]
-
blk.14.attn_output.weightQ5_K[2048, 2048]
-
blk.14.attn_qkv.biasF32[6144]
-
blk.14.attn_qkv.weightQ6_K[2048, 6144]
-
blk.14.ffn_down.weightQ8_0[5504, 2048]
-
blk.14.ffn_gate.weightQ5_K[2048, 5504]
-
blk.14.ffn_norm.weightF32[2048]
-
blk.14.ffn_up.weightQ5_K[2048, 5504]
-
blk.15.attn_norm.weightF32[2048]
-
blk.15.attn_output.weightQ5_K[2048, 2048]
-
blk.15.attn_qkv.biasF32[6144]
-
blk.15.attn_qkv.weightQ6_K[2048, 6144]
-
blk.15.ffn_down.weightQ5_1[5504, 2048]
-
blk.15.ffn_gate.weightQ5_K[2048, 5504]
-
blk.15.ffn_norm.weightF32[2048]
-
blk.15.ffn_up.weightQ5_K[2048, 5504]
-
blk.16.attn_norm.weightF32[2048]
-
blk.16.attn_output.weightQ5_K[2048, 2048]
-
blk.16.attn_qkv.biasF32[6144]
-
blk.16.attn_qkv.weightQ6_K[2048, 6144]
-
blk.16.ffn_down.weightQ5_1[5504, 2048]
-
blk.16.ffn_gate.weightQ5_K[2048, 5504]
-
blk.16.ffn_norm.weightF32[2048]
-
blk.16.ffn_up.weightQ5_K[2048, 5504]
-
blk.17.attn_norm.weightF32[2048]
-
blk.17.attn_output.weightQ5_K[2048, 2048]
-
blk.17.attn_qkv.biasF32[6144]
-
blk.17.attn_qkv.weightQ6_K[2048, 6144]
-
blk.17.ffn_down.weightQ8_0[5504, 2048]
-
blk.17.ffn_gate.weightQ5_K[2048, 5504]
-
blk.17.ffn_norm.weightF32[2048]
-
blk.17.ffn_up.weightQ5_K[2048, 5504]
-
blk.18.attn_norm.weightF32[2048]
-
blk.18.attn_output.weightQ5_K[2048, 2048]
-
blk.18.attn_qkv.biasF32[6144]
-
blk.18.attn_qkv.weightQ6_K[2048, 6144]
-
blk.18.ffn_down.weightQ5_1[5504, 2048]
-
blk.18.ffn_gate.weightQ5_K[2048, 5504]
-
blk.18.ffn_norm.weightF32[2048]
-
blk.18.ffn_up.weightQ5_K[2048, 5504]
-
blk.19.attn_norm.weightF32[2048]
-
blk.19.attn_output.weightQ5_K[2048, 2048]
-
blk.19.attn_qkv.biasF32[6144]
-
blk.19.attn_qkv.weightQ6_K[2048, 6144]
-
blk.19.ffn_down.weightQ5_1[5504, 2048]
-
blk.19.ffn_gate.weightQ5_K[2048, 5504]
-
blk.19.ffn_norm.weightF32[2048]
-
blk.19.ffn_up.weightQ5_K[2048, 5504]
-
blk.20.attn_norm.weightF32[2048]
-
blk.20.attn_output.weightQ5_K[2048, 2048]
-
blk.20.attn_qkv.biasF32[6144]
-
blk.20.attn_qkv.weightQ6_K[2048, 6144]
-
blk.20.ffn_down.weightQ8_0[5504, 2048]
-
blk.20.ffn_gate.weightQ5_K[2048, 5504]
-
blk.20.ffn_norm.weightF32[2048]
-
blk.20.ffn_up.weightQ5_K[2048, 5504]
-
blk.21.attn_norm.weightF32[2048]
-
blk.21.attn_output.weightQ5_K[2048, 2048]
-
blk.21.attn_qkv.biasF32[6144]
-
blk.21.attn_qkv.weightQ6_K[2048, 6144]
-
blk.21.ffn_down.weightQ8_0[5504, 2048]
-
blk.21.ffn_gate.weightQ5_K[2048, 5504]
-
blk.21.ffn_norm.weightF32[2048]
-
blk.21.ffn_up.weightQ5_K[2048, 5504]
-
blk.22.attn_norm.weightF32[2048]
-
blk.22.attn_output.weightQ5_K[2048, 2048]
-
blk.22.attn_qkv.biasF32[6144]
-
blk.22.attn_qkv.weightQ6_K[2048, 6144]
-
blk.22.ffn_down.weightQ8_0[5504, 2048]
-
blk.22.ffn_gate.weightQ5_K[2048, 5504]
-
blk.22.ffn_norm.weightF32[2048]
-
blk.22.ffn_up.weightQ5_K[2048, 5504]
-
blk.23.attn_norm.weightF32[2048]
-
blk.23.attn_output.weightQ5_K[2048, 2048]
-
blk.23.attn_qkv.biasF32[6144]
-
blk.23.attn_qkv.weightQ6_K[2048, 6144]
-
blk.23.ffn_down.weightQ8_0[5504, 2048]
-
blk.23.ffn_gate.weightQ5_K[2048, 5504]
-
blk.23.ffn_norm.weightF32[2048]
-
blk.23.ffn_up.weightQ5_K[2048, 5504]
-
output.weightQ6_K[2048, 151936]
-
output_norm.weightF32[2048]
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
blk.16
blk.17
blk.18
blk.19
blk.20
blk.21
blk.22
blk.23