mitkox/sqlcoder-7b-2-2
119
1{2 "vocab_size": 32016,3 "max_position_embeddings": 16384,4 "hidden_size": 4096,5 "intermediate_size": 11008,6 "num_hidden_layers": 32,7 "num_attention_heads": 32,8 "num_key_value_heads": 32,9 "hidden_act": "silu",10 "initializer_range": 0.02,11 "rms_norm_eps": 1e-05,12 "pretraining_tp": 1,13 "use_cache": true,14 "rope_theta": 1000000,15 "rope_scaling": null,16 "attention_bias": false,17 "attention_dropout": 0.0,18 "return_dict": true,19 "output_hidden_states": false,20 "output_attentions": false,21 "torchscript": false,22 "torch_dtype": "float16",23 "use_bfloat16": false,24 "tf_legacy_loss": false,25 "pruned_heads": {},26 "tie_word_embeddings": false,27 "chunk_size_feed_forward": 0,28 "is_encoder_decoder": false,29 "is_decoder": false,30 "cross_attention_hidden_size": null,31 "add_cross_attention": false,32 "tie_encoder_decoder": false,33 "max_length": 20,34 "min_length": 0,35 "do_sample": false,36 "early_stopping": false,37 "num_beams": 1,38 "num_beam_groups": 1,39 "diversity_penalty": 0.0,40 "temperature": 1.0,41 "top_k": 50,42 "top_p": 1.0,43 "typical_p": 1.0,44 "repetition_penalty": 1.0,45 "length_penalty": 1.0,46 "no_repeat_ngram_size": 0,47 "encoder_no_repeat_ngram_size": 0,48 "bad_words_ids": null,49 "num_return_sequences": 1,50 "output_scores": false,51 "return_dict_in_generate": false,52 "forced_bos_token_id": null,53 "forced_eos_token_id": null,54 "remove_invalid_values": false,55 "exponential_decay_length_penalty": null,56 "suppress_tokens": null,57 "begin_suppress_tokens": null,58 "architectures": [59 "LlamaForCausalLM"60 ],61 "finetuning_task": null,62 "id2label": {63 "0": "LABEL_0",64 "1": "LABEL_1"65 },66 "label2id": {67 "LABEL_0": 0,68 "LABEL_1": 169 },70 "tokenizer_class": null,71 "prefix": null,72 "bos_token_id": 1,73 "pad_token_id": null,74 "eos_token_id": 2,75 "sep_token_id": null,76 "decoder_start_token_id": null,77 "task_specific_params": null,78 "problem_type": null,79 "_name_or_path": "/Users/mitko/.cache/huggingface/hub/models--defog--sqlcoder-7b-2/snapshots/b645fe3146087c7aedea712a900e208b3a3f0152",80 "transformers_version": "4.37.2",81 "model_type": "llama",82 "quantization": {83 "group_size": 64,84 "bits": 485 }86}