1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
| (CodeGeex) ydjsir@YDJ-Z490UD:/mnt/f/model/CodeGeeX2$ python ./demo/run_demo.py --chatglm-cpp fastllm disabled. Using chatglm-cpp to improve performance Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████| 7/7 [00:06<00:00, 1.08it/s]Processing model states: 100%|███████████████████████████████████████████████████████████| 199/199 [00:09<00:00, 21.84it/s]+---------------------------------------------------------------------+---------------------------+---------+ | name | shape | dtype | |---------------------------------------------------------------------+---------------------------+---------| | transformer.embedding.word_embeddings.weight | torch.Size([65024, 4096]) | F16 | | transformer.encoder.layers.0.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.0.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.0.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.0.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.0.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.0.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.0.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.1.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.1.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.1.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.1.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.1.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.1.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.1.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.2.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.2.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.2.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.2.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.2.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.2.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.2.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.3.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.3.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.3.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.3.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.3.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.3.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.3.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.4.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.4.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.4.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.4.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.4.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.4.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.4.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.5.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.5.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.5.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.5.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.5.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.5.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.5.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.6.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.6.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.6.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.6.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.6.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.6.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.6.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.7.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.7.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.7.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.7.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.7.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.7.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.7.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.8.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.8.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.8.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.8.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.8.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.8.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.8.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.9.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.9.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.9.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.9.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.9.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.9.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.9.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.10.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.10.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.10.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.10.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.10.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.10.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.10.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.11.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.11.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.11.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.11.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.11.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.11.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.11.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.12.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.12.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.12.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.12.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.12.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.12.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.12.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.13.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.13.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.13.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.13.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.13.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.13.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.13.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.14.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.14.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.14.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.14.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.14.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.14.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.14.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.15.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.15.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.15.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.15.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.15.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.15.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.15.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.16.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.16.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.16.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.16.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.16.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.16.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.16.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.17.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.17.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.17.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.17.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.17.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.17.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.17.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.18.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.18.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.18.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.18.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.18.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.18.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.18.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.19.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.19.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.19.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.19.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.19.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.19.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.19.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.20.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.20.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.20.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.20.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.20.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.20.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.20.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.21.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.21.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.21.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.21.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.21.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.21.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.21.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.22.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.22.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.22.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.22.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.22.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.22.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.22.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.23.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.23.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.23.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.23.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.23.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.23.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.23.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.24.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.24.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.24.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.24.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.24.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.24.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.24.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.25.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.25.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.25.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.25.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.25.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.25.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.25.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.26.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.26.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.26.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.26.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.26.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.26.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.26.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.layers.27.input_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.27.self_attention.query_key_value.weight | torch.Size([4608, 4096]) | F16 | | transformer.encoder.layers.27.self_attention.query_key_value.bias | torch.Size([4608]) | F32 | | transformer.encoder.layers.27.self_attention.dense.weight | torch.Size([4096, 4096]) | F16 | | transformer.encoder.layers.27.post_attention_layernorm.weight | torch.Size([4096]) | F32 | | transformer.encoder.layers.27.mlp.dense_h_to_4h.weight | torch.Size([27392, 4096]) | F16 | | transformer.encoder.layers.27.mlp.dense_4h_to_h.weight | torch.Size([4096, 13696]) | F16 | | transformer.encoder.final_layernorm.weight | torch.Size([4096]) | F32 | | transformer.output_layer.weight | torch.Size([65024, 4096]) | F16 | +---------------------------------------------------------------------+---------------------------+---------+ ggml_init_cublas: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 2080 Ti, compute capability 7.5 Running on local URL: http://0.0.0.0:7861
To create a public link, set `share=True` in `launch()`.
|