Update config.json
Browse files- config.json +0 -84
config.json
CHANGED
|
@@ -42,90 +42,6 @@
|
|
| 42 |
"LABEL_1": 1
|
| 43 |
},
|
| 44 |
"length_penalty": 1.0,
|
| 45 |
-
"llm_config": {
|
| 46 |
-
"vocab_size": 151936,
|
| 47 |
-
"max_position_embeddings": 32768,
|
| 48 |
-
"hidden_size": 4096,
|
| 49 |
-
"intermediate_size": 22016,
|
| 50 |
-
"num_hidden_layers": 32,
|
| 51 |
-
"num_attention_heads": 32,
|
| 52 |
-
"use_sliding_window": false,
|
| 53 |
-
"sliding_window": 4096,
|
| 54 |
-
"max_window_layers": 28,
|
| 55 |
-
"num_key_value_heads": 32,
|
| 56 |
-
"hidden_act": "silu",
|
| 57 |
-
"initializer_range": 0.02,
|
| 58 |
-
"rms_norm_eps": 1e-06,
|
| 59 |
-
"use_cache": true,
|
| 60 |
-
"rope_theta": 10000.0,
|
| 61 |
-
"rope_scaling": null,
|
| 62 |
-
"attention_dropout": 0.0,
|
| 63 |
-
"return_dict": true,
|
| 64 |
-
"output_hidden_states": false,
|
| 65 |
-
"output_attentions": false,
|
| 66 |
-
"torchscript": false,
|
| 67 |
-
"torch_dtype": null,
|
| 68 |
-
"use_bfloat16": false,
|
| 69 |
-
"tf_legacy_loss": false,
|
| 70 |
-
"pruned_heads": {},
|
| 71 |
-
"tie_word_embeddings": false,
|
| 72 |
-
"chunk_size_feed_forward": 0,
|
| 73 |
-
"is_encoder_decoder": false,
|
| 74 |
-
"is_decoder": false,
|
| 75 |
-
"cross_attention_hidden_size": null,
|
| 76 |
-
"add_cross_attention": false,
|
| 77 |
-
"tie_encoder_decoder": false,
|
| 78 |
-
"max_length": 20,
|
| 79 |
-
"min_length": 0,
|
| 80 |
-
"do_sample": false,
|
| 81 |
-
"early_stopping": false,
|
| 82 |
-
"num_beams": 1,
|
| 83 |
-
"num_beam_groups": 1,
|
| 84 |
-
"diversity_penalty": 0.0,
|
| 85 |
-
"temperature": 1.0,
|
| 86 |
-
"top_k": 50,
|
| 87 |
-
"top_p": 1.0,
|
| 88 |
-
"typical_p": 1.0,
|
| 89 |
-
"repetition_penalty": 1.0,
|
| 90 |
-
"length_penalty": 1.0,
|
| 91 |
-
"no_repeat_ngram_size": 0,
|
| 92 |
-
"encoder_no_repeat_ngram_size": 0,
|
| 93 |
-
"bad_words_ids": null,
|
| 94 |
-
"num_return_sequences": 1,
|
| 95 |
-
"output_scores": false,
|
| 96 |
-
"return_dict_in_generate": false,
|
| 97 |
-
"forced_bos_token_id": null,
|
| 98 |
-
"forced_eos_token_id": null,
|
| 99 |
-
"remove_invalid_values": false,
|
| 100 |
-
"exponential_decay_length_penalty": null,
|
| 101 |
-
"suppress_tokens": null,
|
| 102 |
-
"begin_suppress_tokens": null,
|
| 103 |
-
"architectures": [
|
| 104 |
-
"Qwen2ForCausalLM"
|
| 105 |
-
],
|
| 106 |
-
"finetuning_task": null,
|
| 107 |
-
"id2label": {
|
| 108 |
-
"0": "LABEL_0",
|
| 109 |
-
"1": "LABEL_1"
|
| 110 |
-
},
|
| 111 |
-
"label2id": {
|
| 112 |
-
"LABEL_0": 0,
|
| 113 |
-
"LABEL_1": 1
|
| 114 |
-
},
|
| 115 |
-
"tokenizer_class": null,
|
| 116 |
-
"prefix": null,
|
| 117 |
-
"bos_token_id": null,
|
| 118 |
-
"pad_token_id": null,
|
| 119 |
-
"eos_token_id": null,
|
| 120 |
-
"sep_token_id": null,
|
| 121 |
-
"decoder_start_token_id": null,
|
| 122 |
-
"task_specific_params": null,
|
| 123 |
-
"problem_type": null,
|
| 124 |
-
"_name_or_path": "",
|
| 125 |
-
"_attn_implementation_autoset": false,
|
| 126 |
-
"transformers_version": "4.51.3",
|
| 127 |
-
"model_type": "qwen2"
|
| 128 |
-
},
|
| 129 |
"max_dynamic_patch": 12,
|
| 130 |
"max_length": 20,
|
| 131 |
"min_dynamic_patch": 1,
|
|
|
|
| 42 |
"LABEL_1": 1
|
| 43 |
},
|
| 44 |
"length_penalty": 1.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
"max_dynamic_patch": 12,
|
| 46 |
"max_length": 20,
|
| 47 |
"min_dynamic_patch": 1,
|