| { | |
| "_name_or_path": "distilbert/distilbert-base-uncased", | |
| "activation": "gelu", | |
| "architectures": [ | |
| "DistilBertForSequenceClassification" | |
| ], | |
| "attention_dropout": 0.1, | |
| "dim": 768, | |
| "dropout": 0.1, | |
| "hidden_dim": 3072, | |
| "id2label": { | |
| "0": "Bloom-7B", | |
| "1": "Claude-Instant-v1", | |
| "2": "Claude-v1", | |
| "3": "Cohere-Command", | |
| "4": "Dolphin-2.5-Mixtral-8x7B", | |
| "5": "Dolphin-Mixtral-8x7B", | |
| "6": "Falcon-180B", | |
| "7": "Flan-T5-Base", | |
| "8": "Flan-T5-Large", | |
| "9": "Flan-T5-Small", | |
| "10": "Flan-T5-XL", | |
| "11": "Flan-T5-XXL", | |
| "12": "GLM-130B", | |
| "13": "GPT-3.5", | |
| "14": "GPT-4", | |
| "15": "GPT-J", | |
| "16": "GPT-NeoX", | |
| "17": "Gemini-Pro", | |
| "18": "Goliath-120B", | |
| "19": "Human", | |
| "20": "LLaMA-13B", | |
| "21": "LLaMA-2-70B", | |
| "22": "LLaMA-2-7B", | |
| "23": "LLaMA-30B", | |
| "24": "LLaMA-65B", | |
| "25": "LLaMA-7B", | |
| "26": "LZLV-70B", | |
| "27": "Mistral-7B", | |
| "28": "Mistral-7B-OpenOrca", | |
| "29": "Mixtral-8x7B", | |
| "30": "MythoMax-L2-13B", | |
| "31": "Neural-Chat-7B", | |
| "32": "Noromaid-20B", | |
| "33": "Nous-Capybara-34B", | |
| "34": "Nous-Capybara-7B", | |
| "35": "Nous-Hermes-LLaMA-2-13B", | |
| "36": "Nous-Hermes-LLaMA-2-70B", | |
| "37": "OPT-1.3B", | |
| "38": "OPT-125M", | |
| "39": "OPT-13B", | |
| "40": "OPT-2.7B", | |
| "41": "OPT-30B", | |
| "42": "OPT-350M", | |
| "43": "OPT-6.7B", | |
| "44": "OpenChat-3.5", | |
| "45": "OpenHermes-2-Mistral-7B", | |
| "46": "OpenHermes-2.5-Mistral-7B", | |
| "47": "PaLM-2", | |
| "48": "Psyfighter-13B", | |
| "49": "Psyfighter-2-13B", | |
| "50": "RWKV-5-World-3B", | |
| "51": "StripedHyena-Nous-7B", | |
| "52": "T0-11B", | |
| "53": "T0-3B", | |
| "54": "Text-Ada-001", | |
| "55": "Text-Babbage-001", | |
| "56": "Text-Curie-001", | |
| "57": "Text-Davinci-001", | |
| "58": "Text-Davinci-002", | |
| "59": "Text-Davinci-003", | |
| "60": "Toppy-M-7B", | |
| "61": "Unknown", | |
| "62": "YI-34B" | |
| }, | |
| "initializer_range": 0.02, | |
| "label2id": { | |
| "Bloom-7B": 0, | |
| "Claude-Instant-v1": 1, | |
| "Claude-v1": 2, | |
| "Cohere-Command": 3, | |
| "Dolphin-2.5-Mixtral-8x7B": 4, | |
| "Dolphin-Mixtral-8x7B": 5, | |
| "Falcon-180B": 6, | |
| "Flan-T5-Base": 7, | |
| "Flan-T5-Large": 8, | |
| "Flan-T5-Small": 9, | |
| "Flan-T5-XL": 10, | |
| "Flan-T5-XXL": 11, | |
| "GLM-130B": 12, | |
| "GPT-3.5": 13, | |
| "GPT-4": 14, | |
| "GPT-J": 15, | |
| "GPT-NeoX": 16, | |
| "Gemini-Pro": 17, | |
| "Goliath-120B": 18, | |
| "Human": 19, | |
| "LLaMA-13B": 20, | |
| "LLaMA-2-70B": 21, | |
| "LLaMA-2-7B": 22, | |
| "LLaMA-30B": 23, | |
| "LLaMA-65B": 24, | |
| "LLaMA-7B": 25, | |
| "LZLV-70B": 26, | |
| "Mistral-7B": 27, | |
| "Mistral-7B-OpenOrca": 28, | |
| "Mixtral-8x7B": 29, | |
| "MythoMax-L2-13B": 30, | |
| "Neural-Chat-7B": 31, | |
| "Noromaid-20B": 32, | |
| "Nous-Capybara-34B": 33, | |
| "Nous-Capybara-7B": 34, | |
| "Nous-Hermes-LLaMA-2-13B": 35, | |
| "Nous-Hermes-LLaMA-2-70B": 36, | |
| "OPT-1.3B": 37, | |
| "OPT-125M": 38, | |
| "OPT-13B": 39, | |
| "OPT-2.7B": 40, | |
| "OPT-30B": 41, | |
| "OPT-350M": 42, | |
| "OPT-6.7B": 43, | |
| "OpenChat-3.5": 44, | |
| "OpenHermes-2-Mistral-7B": 45, | |
| "OpenHermes-2.5-Mistral-7B": 46, | |
| "PaLM-2": 47, | |
| "Psyfighter-13B": 48, | |
| "Psyfighter-2-13B": 49, | |
| "RWKV-5-World-3B": 50, | |
| "StripedHyena-Nous-7B": 51, | |
| "T0-11B": 52, | |
| "T0-3B": 53, | |
| "Text-Ada-001": 54, | |
| "Text-Babbage-001": 55, | |
| "Text-Curie-001": 56, | |
| "Text-Davinci-001": 57, | |
| "Text-Davinci-002": 58, | |
| "Text-Davinci-003": 59, | |
| "Toppy-M-7B": 60, | |
| "Unknown": 61, | |
| "YI-34B": 62 | |
| }, | |
| "max_position_embeddings": 512, | |
| "model_type": "distilbert", | |
| "n_heads": 12, | |
| "n_layers": 6, | |
| "pad_token_id": 0, | |
| "qa_dropout": 0.1, | |
| "seq_classif_dropout": 0.2, | |
| "sinusoidal_pos_embds": false, | |
| "tie_weights_": true, | |
| "transformers_version": "4.40.1", | |
| "vocab_size": 30522 | |
| } | |