gtonkov/SpecAI

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": "configuration_phi3.Phi3Config",10    "AutoModel": "modeling_phi3.Phi3ForCausalLM",11    "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"12  },13  "bos_token_id": 1,14  "embd_pdrop": 0.0,15  "eos_token_id": 32000,16  "hidden_act": "silu",17  "hidden_size": 3072,18  "initializer_range": 0.02,19  "intermediate_size": 8192,20  "max_position_embeddings": 131072,21  "model_type": "phi3",22  "num_attention_heads": 32,23  "num_hidden_layers": 32,24  "num_key_value_heads": 32,25  "original_max_position_embeddings": 4096,26  "pad_token_id": 32000,27  "resid_pdrop": 0.0,28  "rms_norm_eps": 1e-05,29  "rope_scaling": {30    "long_factor": [31      1.0800000429153442,32      1.1100000143051147,33      1.1399999856948853,34      1.340000033378601,35      1.5899999141693115,36      1.600000023841858,37      1.6200000047683716,38      2.620000123977661,39      3.2300000190734863,40      3.2300000190734863,41      4.789999961853027,42      7.400000095367432,43      7.700000286102295,44      9.09000015258789,45      12.199999809265137,46      17.670000076293945,47      24.46000099182129,48      28.57000160217285,49      30.420001983642578,50      30.840002059936523,51      32.590003967285156,52      32.93000411987305,53      42.320003509521484,54      44.96000289916992,55      50.340003967285156,56      50.45000457763672,57      57.55000305175781,58      57.93000411987305,59      58.21000289916992,60      60.1400032043457,61      62.61000442504883,62      62.62000274658203,63      62.71000289916992,64      63.1400032043457,65      63.1400032043457,66      63.77000427246094,67      63.93000411987305,68      63.96000289916992,69      63.970001220703125,70      64.02999877929688,71      64.06999969482422,72      64.08000183105469,73      64.12000274658203,74      64.41000366210938,75      64.4800033569336,76      64.51000213623047,77      64.52999877929688,78      64.8399963378906279    ],80    "short_factor": [81      1.0,82      1.0199999809265137,83      1.0299999713897705,84      1.0299999713897705,85      1.0499999523162842,86      1.0499999523162842,87      1.0499999523162842,88      1.0499999523162842,89      1.0499999523162842,90      1.0699999332427979,91      1.0999999046325684,92      1.1099998950958252,93      1.1599998474121094,94      1.1599998474121094,95      1.1699998378753662,96      1.2899998426437378,97      1.339999794960022,98      1.679999828338623,99      1.7899998426437378,100      1.8199998140335083,101      1.8499997854232788,102      1.8799997568130493,103      1.9099997282028198,104      1.9399996995925903,105      1.9899996519088745,106      2.0199997425079346,107      2.0199997425079346,108      2.0199997425079346,109      2.0199997425079346,110      2.0199997425079346,111      2.0199997425079346,112      2.0299997329711914,113      2.0299997329711914,114      2.0299997329711914,115      2.0299997329711914,116      2.0299997329711914,117      2.0299997329711914,118      2.0299997329711914,119      2.0299997329711914,120      2.0299997329711914,121      2.0799996852874756,122      2.0899996757507324,123      2.189999580383301,124      2.2199995517730713,125      2.5899994373321533,126      2.729999542236328,127      2.749999523162842,128      2.8399994373321533129    ],130    "type": "longrope"131  },132  "rope_theta": 10000.0,133  "sliding_window": 262144,134  "tie_word_embeddings": false,135  "torch_dtype": "bfloat16",136  "transformers_version": "4.44.2",137  "use_cache": false,138  "vocab_size": 32064139}140