Add config.json
Browse files- mlx/llm-dwq4/config.json +89 -0
mlx/llm-dwq4/config.json
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_cross_attention": false,
|
| 3 |
+
"architectures": [
|
| 4 |
+
"Qwen3ForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 151643,
|
| 9 |
+
"chunk_size_feed_forward": 0,
|
| 10 |
+
"diversity_penalty": 0.0,
|
| 11 |
+
"do_sample": false,
|
| 12 |
+
"early_stopping": false,
|
| 13 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 14 |
+
"eos_token_id": [
|
| 15 |
+
151643,
|
| 16 |
+
151645
|
| 17 |
+
],
|
| 18 |
+
"head_dim": 128,
|
| 19 |
+
"hidden_act": "silu",
|
| 20 |
+
"hidden_size": 2048,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 6144,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"length_penalty": 1.0,
|
| 34 |
+
"max_length": 20,
|
| 35 |
+
"max_position_embeddings": 65536,
|
| 36 |
+
"min_length": 0,
|
| 37 |
+
"model_type": "qwen3",
|
| 38 |
+
"no_repeat_ngram_size": 0,
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_beam_groups": 1,
|
| 41 |
+
"num_beams": 1,
|
| 42 |
+
"num_hidden_layers": 28,
|
| 43 |
+
"num_key_value_heads": 8,
|
| 44 |
+
"num_return_sequences": 1,
|
| 45 |
+
"output_attentions": false,
|
| 46 |
+
"output_hidden_states": false,
|
| 47 |
+
"output_scores": false,
|
| 48 |
+
"pad_token_id": 151643,
|
| 49 |
+
"pruned_heads": {},
|
| 50 |
+
"quantization": {
|
| 51 |
+
"group_size": 64,
|
| 52 |
+
"bits": 4,
|
| 53 |
+
"mode": "affine"
|
| 54 |
+
},
|
| 55 |
+
"quantization_config": {
|
| 56 |
+
"group_size": 64,
|
| 57 |
+
"bits": 4,
|
| 58 |
+
"mode": "affine"
|
| 59 |
+
},
|
| 60 |
+
"remove_invalid_values": false,
|
| 61 |
+
"repetition_penalty": 1.0,
|
| 62 |
+
"return_dict": true,
|
| 63 |
+
"return_dict_in_generate": false,
|
| 64 |
+
"rms_norm_eps": 1e-06,
|
| 65 |
+
"rope_scaling": {
|
| 66 |
+
"interleaved": true,
|
| 67 |
+
"mrope_interleaved": true,
|
| 68 |
+
"mrope_section": [
|
| 69 |
+
24,
|
| 70 |
+
20,
|
| 71 |
+
20
|
| 72 |
+
],
|
| 73 |
+
"rope_type": "default",
|
| 74 |
+
"type": "default"
|
| 75 |
+
},
|
| 76 |
+
"rope_theta": 1000000,
|
| 77 |
+
"temperature": 1.0,
|
| 78 |
+
"tf_legacy_loss": false,
|
| 79 |
+
"tie_encoder_decoder": false,
|
| 80 |
+
"tie_word_embeddings": true,
|
| 81 |
+
"top_k": 50,
|
| 82 |
+
"top_p": 1.0,
|
| 83 |
+
"torch_dtype": "bfloat16",
|
| 84 |
+
"torchscript": false,
|
| 85 |
+
"typical_p": 1.0,
|
| 86 |
+
"use_bfloat16": false,
|
| 87 |
+
"use_cache": true,
|
| 88 |
+
"vocab_size": 151936
|
| 89 |
+
}
|