Upload config.json with huggingface_hub
Browse files- config.json +138 -0
    	
        config.json
    ADDED
    
    | @@ -0,0 +1,138 @@ | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | 
|  | |
| 1 | 
            +
            {
         | 
| 2 | 
            +
              "_name_or_path": "microsoft/Phi-3.5-mini-instruct",
         | 
| 3 | 
            +
              "architectures": [
         | 
| 4 | 
            +
                "Phi3ForCausalLM"
         | 
| 5 | 
            +
              ],
         | 
| 6 | 
            +
              "attention_bias": false,
         | 
| 7 | 
            +
              "attention_dropout": 0.0,
         | 
| 8 | 
            +
              "auto_map": {
         | 
| 9 | 
            +
                "AutoConfig": "microsoft/Phi-3.5-mini-instruct--configuration_phi3.Phi3Config",
         | 
| 10 | 
            +
                "AutoModelForCausalLM": "microsoft/Phi-3.5-mini-instruct--modeling_phi3.Phi3ForCausalLM"
         | 
| 11 | 
            +
              },
         | 
| 12 | 
            +
              "bos_token_id": 1,
         | 
| 13 | 
            +
              "embd_pdrop": 0.0,
         | 
| 14 | 
            +
              "eos_token_id": 32000,
         | 
| 15 | 
            +
              "hidden_act": "silu",
         | 
| 16 | 
            +
              "hidden_size": 3072,
         | 
| 17 | 
            +
              "initializer_range": 0.02,
         | 
| 18 | 
            +
              "intermediate_size": 8192,
         | 
| 19 | 
            +
              "max_position_embeddings": 4096,
         | 
| 20 | 
            +
              "model_type": "phi3",
         | 
| 21 | 
            +
              "num_attention_heads": 32,
         | 
| 22 | 
            +
              "num_hidden_layers": 32,
         | 
| 23 | 
            +
              "num_key_value_heads": 32,
         | 
| 24 | 
            +
              "original_max_position_embeddings": 4096,
         | 
| 25 | 
            +
              "pad_token_id": 32000,
         | 
| 26 | 
            +
              "resid_pdrop": 0.0,
         | 
| 27 | 
            +
              "rms_norm_eps": 1e-05,
         | 
| 28 | 
            +
              "rope_scaling": {
         | 
| 29 | 
            +
                "long_factor": [
         | 
| 30 | 
            +
                  1.0800000429153442,
         | 
| 31 | 
            +
                  1.1100000143051147,
         | 
| 32 | 
            +
                  1.1399999856948853,
         | 
| 33 | 
            +
                  1.340000033378601,
         | 
| 34 | 
            +
                  1.5899999141693115,
         | 
| 35 | 
            +
                  1.600000023841858,
         | 
| 36 | 
            +
                  1.6200000047683716,
         | 
| 37 | 
            +
                  2.620000123977661,
         | 
| 38 | 
            +
                  3.2300000190734863,
         | 
| 39 | 
            +
                  3.2300000190734863,
         | 
| 40 | 
            +
                  4.789999961853027,
         | 
| 41 | 
            +
                  7.400000095367432,
         | 
| 42 | 
            +
                  7.700000286102295,
         | 
| 43 | 
            +
                  9.09000015258789,
         | 
| 44 | 
            +
                  12.199999809265137,
         | 
| 45 | 
            +
                  17.670000076293945,
         | 
| 46 | 
            +
                  24.46000099182129,
         | 
| 47 | 
            +
                  28.57000160217285,
         | 
| 48 | 
            +
                  30.420001983642578,
         | 
| 49 | 
            +
                  30.840002059936523,
         | 
| 50 | 
            +
                  32.590003967285156,
         | 
| 51 | 
            +
                  32.93000411987305,
         | 
| 52 | 
            +
                  42.320003509521484,
         | 
| 53 | 
            +
                  44.96000289916992,
         | 
| 54 | 
            +
                  50.340003967285156,
         | 
| 55 | 
            +
                  50.45000457763672,
         | 
| 56 | 
            +
                  57.55000305175781,
         | 
| 57 | 
            +
                  57.93000411987305,
         | 
| 58 | 
            +
                  58.21000289916992,
         | 
| 59 | 
            +
                  60.1400032043457,
         | 
| 60 | 
            +
                  62.61000442504883,
         | 
| 61 | 
            +
                  62.62000274658203,
         | 
| 62 | 
            +
                  62.71000289916992,
         | 
| 63 | 
            +
                  63.1400032043457,
         | 
| 64 | 
            +
                  63.1400032043457,
         | 
| 65 | 
            +
                  63.77000427246094,
         | 
| 66 | 
            +
                  63.93000411987305,
         | 
| 67 | 
            +
                  63.96000289916992,
         | 
| 68 | 
            +
                  63.970001220703125,
         | 
| 69 | 
            +
                  64.02999877929688,
         | 
| 70 | 
            +
                  64.06999969482422,
         | 
| 71 | 
            +
                  64.08000183105469,
         | 
| 72 | 
            +
                  64.12000274658203,
         | 
| 73 | 
            +
                  64.41000366210938,
         | 
| 74 | 
            +
                  64.4800033569336,
         | 
| 75 | 
            +
                  64.51000213623047,
         | 
| 76 | 
            +
                  64.52999877929688,
         | 
| 77 | 
            +
                  64.83999633789062
         | 
| 78 | 
            +
                ],
         | 
| 79 | 
            +
                "short_factor": [
         | 
| 80 | 
            +
                  1.0,
         | 
| 81 | 
            +
                  1.0199999809265137,
         | 
| 82 | 
            +
                  1.0299999713897705,
         | 
| 83 | 
            +
                  1.0299999713897705,
         | 
| 84 | 
            +
                  1.0499999523162842,
         | 
| 85 | 
            +
                  1.0499999523162842,
         | 
| 86 | 
            +
                  1.0499999523162842,
         | 
| 87 | 
            +
                  1.0499999523162842,
         | 
| 88 | 
            +
                  1.0499999523162842,
         | 
| 89 | 
            +
                  1.0699999332427979,
         | 
| 90 | 
            +
                  1.0999999046325684,
         | 
| 91 | 
            +
                  1.1099998950958252,
         | 
| 92 | 
            +
                  1.1599998474121094,
         | 
| 93 | 
            +
                  1.1599998474121094,
         | 
| 94 | 
            +
                  1.1699998378753662,
         | 
| 95 | 
            +
                  1.2899998426437378,
         | 
| 96 | 
            +
                  1.339999794960022,
         | 
| 97 | 
            +
                  1.679999828338623,
         | 
| 98 | 
            +
                  1.7899998426437378,
         | 
| 99 | 
            +
                  1.8199998140335083,
         | 
| 100 | 
            +
                  1.8499997854232788,
         | 
| 101 | 
            +
                  1.8799997568130493,
         | 
| 102 | 
            +
                  1.9099997282028198,
         | 
| 103 | 
            +
                  1.9399996995925903,
         | 
| 104 | 
            +
                  1.9899996519088745,
         | 
| 105 | 
            +
                  2.0199997425079346,
         | 
| 106 | 
            +
                  2.0199997425079346,
         | 
| 107 | 
            +
                  2.0199997425079346,
         | 
| 108 | 
            +
                  2.0199997425079346,
         | 
| 109 | 
            +
                  2.0199997425079346,
         | 
| 110 | 
            +
                  2.0199997425079346,
         | 
| 111 | 
            +
                  2.0299997329711914,
         | 
| 112 | 
            +
                  2.0299997329711914,
         | 
| 113 | 
            +
                  2.0299997329711914,
         | 
| 114 | 
            +
                  2.0299997329711914,
         | 
| 115 | 
            +
                  2.0299997329711914,
         | 
| 116 | 
            +
                  2.0299997329711914,
         | 
| 117 | 
            +
                  2.0299997329711914,
         | 
| 118 | 
            +
                  2.0299997329711914,
         | 
| 119 | 
            +
                  2.0299997329711914,
         | 
| 120 | 
            +
                  2.0799996852874756,
         | 
| 121 | 
            +
                  2.0899996757507324,
         | 
| 122 | 
            +
                  2.189999580383301,
         | 
| 123 | 
            +
                  2.2199995517730713,
         | 
| 124 | 
            +
                  2.5899994373321533,
         | 
| 125 | 
            +
                  2.729999542236328,
         | 
| 126 | 
            +
                  2.749999523162842,
         | 
| 127 | 
            +
                  2.8399994373321533
         | 
| 128 | 
            +
                ],
         | 
| 129 | 
            +
                "type": "longrope"
         | 
| 130 | 
            +
              },
         | 
| 131 | 
            +
              "rope_theta": 10000.0,
         | 
| 132 | 
            +
              "sliding_window": 262144,
         | 
| 133 | 
            +
              "tie_word_embeddings": false,
         | 
| 134 | 
            +
              "torch_dtype": "bfloat16",
         | 
| 135 | 
            +
              "transformers_version": "4.45.2",
         | 
| 136 | 
            +
              "use_cache": true,
         | 
| 137 | 
            +
              "vocab_size": 32064
         | 
| 138 | 
            +
            }
         | 
