cahaj/Phi-3.5-mini-instruct-text2sql

1{2  "_name_or_path": "unsloth/Phi-3.5-mini-instruct",3  "architectures": [4    "LlamaForCausalLM"5  ],6  "attention_bias": false,7  "attention_dropout": 0.0,8  "bos_token_id": 1,9  "eos_token_id": 32000,10  "hidden_act": "silu",11  "hidden_size": 3072,12  "initializer_range": 0.02,13  "intermediate_size": 8192,14  "max_position_embeddings": 131072,15  "mlp_bias": false,16  "model_type": "llama",17  "num_attention_heads": 32,18  "num_hidden_layers": 32,19  "num_key_value_heads": 32,20  "original_max_position_embeddings": 4096,21  "pad_token_id": 32000,22  "pretraining_tp": 1,23  "rms_norm_eps": 1e-05,24  "rope_scaling": {25    "attention_factor": 32.0,26    "factor": 32.0,27    "long_factor": [28      1.0800000429153442,29      1.1100000143051147,30      1.1399999856948853,31      1.340000033378601,32      1.5899999141693115,33      1.600000023841858,34      1.6200000047683716,35      2.620000123977661,36      3.2300000190734863,37      3.2300000190734863,38      4.789999961853027,39      7.400000095367432,40      7.700000286102295,41      9.09000015258789,42      12.199999809265137,43      17.670000076293945,44      24.46000099182129,45      28.57000160217285,46      30.420001983642578,47      30.840002059936523,48      32.590003967285156,49      32.93000411987305,50      42.320003509521484,51      44.96000289916992,52      50.340003967285156,53      50.45000457763672,54      57.55000305175781,55      57.93000411987305,56      58.21000289916992,57      60.1400032043457,58      62.61000442504883,59      62.62000274658203,60      62.71000289916992,61      63.1400032043457,62      63.1400032043457,63      63.77000427246094,64      63.93000411987305,65      63.96000289916992,66      63.970001220703125,67      64.02999877929688,68      64.06999969482422,69      64.08000183105469,70      64.12000274658203,71      64.41000366210938,72      64.4800033569336,73      64.51000213623047,74      64.52999877929688,75      64.8399963378906276    ],77    "rope_type": "longrope",78    "short_factor": [79      1.0,80      1.0199999809265137,81      1.0299999713897705,82      1.0299999713897705,83      1.0499999523162842,84      1.0499999523162842,85      1.0499999523162842,86      1.0499999523162842,87      1.0499999523162842,88      1.0699999332427979,89      1.0999999046325684,90      1.1099998950958252,91      1.1599998474121094,92      1.1599998474121094,93      1.1699998378753662,94      1.2899998426437378,95      1.339999794960022,96      1.679999828338623,97      1.7899998426437378,98      1.8199998140335083,99      1.8499997854232788,100      1.8799997568130493,101      1.9099997282028198,102      1.9399996995925903,103      1.9899996519088745,104      2.0199997425079346,105      2.0199997425079346,106      2.0199997425079346,107      2.0199997425079346,108      2.0199997425079346,109      2.0199997425079346,110      2.0299997329711914,111      2.0299997329711914,112      2.0299997329711914,113      2.0299997329711914,114      2.0299997329711914,115      2.0299997329711914,116      2.0299997329711914,117      2.0299997329711914,118      2.0299997329711914,119      2.0799996852874756,120      2.0899996757507324,121      2.189999580383301,122      2.2199995517730713,123      2.5899994373321533,124      2.729999542236328,125      2.749999523162842,126      2.8399994373321533127    ]128  },129  "rope_theta": 10000.0,130  "tie_word_embeddings": false,131  "torch_dtype": "bfloat16",132  "transformers_version": "4.44.2",133  "unsloth_version": "2024.8",134  "use_cache": true,135  "vocab_size": 32064136}137