Mini-Bleyz/Bleyzos-Coder

1{
2  "architectures": [
3    "MiMoV2ForCausalLM"
4  ],
5  "auto_map": {
6    "AutoConfig": "configuration_mimo_v2.MiMoV2Config",
7    "AutoModel": "modeling_mimo_v2.MiMoV2Model",
8    "AutoModelForCausalLM": "modeling_mimo_v2.MiMoV2ForCausalLM"
9  },
10  "quantization_config": {
11    "activation_scheme": "dynamic",
12    "fmt": "e4m3",
13    "quant_method": "fp8",
14    "weight_block_size": [
15      128,
16      128
17    ],
18    "ignored_layers": [
19      "model.layers.0.self_attn.o_proj",
20      "model.layers.1.self_attn.o_proj",
21      "model.layers.2.self_attn.o_proj",
22      "model.layers.3.self_attn.o_proj",
23      "model.layers.4.self_attn.o_proj",
24      "model.layers.5.self_attn.o_proj",
25      "model.layers.6.self_attn.o_proj",
26      "model.layers.7.self_attn.o_proj",
27      "model.layers.8.self_attn.o_proj",
28      "model.layers.9.self_attn.o_proj",
29      "model.layers.10.self_attn.o_proj",
30      "model.layers.11.self_attn.o_proj",
31      "model.layers.12.self_attn.o_proj",
32      "model.layers.13.self_attn.o_proj",
33      "model.layers.14.self_attn.o_proj",
34      "model.layers.15.self_attn.o_proj",
35      "model.layers.16.self_attn.o_proj",
36      "model.layers.17.self_attn.o_proj",
37      "model.layers.18.self_attn.o_proj",
38      "model.layers.19.self_attn.o_proj",
39      "model.layers.20.self_attn.o_proj",
40      "model.layers.21.self_attn.o_proj",
41      "model.layers.22.self_attn.o_proj",
42      "model.layers.23.self_attn.o_proj",
43      "model.layers.24.self_attn.o_proj",
44      "model.layers.25.self_attn.o_proj",
45      "model.layers.26.self_attn.o_proj",
46      "model.layers.27.self_attn.o_proj",
47      "model.layers.28.self_attn.o_proj",
48      "model.layers.29.self_attn.o_proj",
49      "model.layers.30.self_attn.o_proj",
50      "model.layers.31.self_attn.o_proj",
51      "model.layers.32.self_attn.o_proj",
52      "model.layers.33.self_attn.o_proj",
53      "model.layers.34.self_attn.o_proj",
54      "model.layers.35.self_attn.o_proj",
55      "model.layers.36.self_attn.o_proj",
56      "model.layers.37.self_attn.o_proj",
57      "model.layers.38.self_attn.o_proj",
58      "model.layers.39.self_attn.o_proj",
59      "model.layers.40.self_attn.o_proj",
60      "model.layers.41.self_attn.o_proj",
61      "model.layers.42.self_attn.o_proj",
62      "model.layers.43.self_attn.o_proj",
63      "model.layers.44.self_attn.o_proj",
64      "model.layers.45.self_attn.o_proj",
65      "model.layers.46.self_attn.o_proj",
66      "model.layers.47.self_attn.o_proj",
67      "model.layers.48.self_attn.o_proj",
68      "model.layers.49.self_attn.o_proj",
69      "model.layers.50.self_attn.o_proj",
70      "model.layers.51.self_attn.o_proj",
71      "model.layers.52.self_attn.o_proj",
72      "model.layers.53.self_attn.o_proj",
73      "model.layers.54.self_attn.o_proj",
74      "model.layers.55.self_attn.o_proj",
75      "model.layers.56.self_attn.o_proj",
76      "model.layers.57.self_attn.o_proj",
77      "model.layers.58.self_attn.o_proj",
78      "model.layers.59.self_attn.o_proj",
79      "model.layers.60.self_attn.o_proj",
80      "model.layers.61.self_attn.o_proj",
81      "model.layers.62.self_attn.o_proj",
82      "model.layers.63.self_attn.o_proj",
83      "model.layers.64.self_attn.o_proj",
84      "model.layers.65.self_attn.o_proj",
85      "model.layers.66.self_attn.o_proj",
86      "model.layers.67.self_attn.o_proj",
87      "model.layers.68.self_attn.o_proj",
88      "model.layers.69.self_attn.o_proj",
89      "model.decoder.self_attn.o_proj"
90    ]
91  },
92  "add_full_attention_sink_bias": false,
93  "add_swa_attention_sink_bias": true,
94  "attention_bias": false,
95  "attention_chunk_size": 128,
96  "attention_dropout": 0.0,
97  "attention_projection_layout": "fused_qkv",
98  "attention_value_scale": 0.612,
99  "head_dim": 192,
100  "hidden_act": "silu",
101  "hidden_size": 6144,
102  "hybrid_layer_pattern": [
103    0, 1, 1, 1, 1, 1, 1,
104    0, 1, 1, 1, 1, 1, 1, 1,
105    0, 1, 1, 1, 1, 1, 1, 1,
106    0, 1, 1, 1, 1, 1, 1, 1,
107    0, 1, 1, 1, 1, 1, 1, 1,
108    0, 1, 1, 1, 1, 1, 1, 1,
109    0, 1, 1, 1, 1, 1, 1, 1,
110    0, 1, 1, 1, 1, 1, 1,
111    0, 1, 1, 1, 1, 1, 1,
112    0
113  ],
114  "initializer_range": 0.02,
115  "intermediate_size": 16384,
116  "layernorm_epsilon": 1e-05,
117  "max_position_embeddings": 1048576,
118  "model_type": "mimo_v2",
119  "moe_intermediate_size": 2048,
120  "moe_layer_freq": [
121    0, 1, 1, 1, 1, 1, 1, 1, 1, 1,
122    1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
123    1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
124    1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
125    1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
126    1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
127    1, 1, 1, 1, 1, 1, 1, 1, 1, 1
128  ],
129  "n_group": 1,
130  "n_routed_experts": 384,
131  "n_shared_experts": null,
132  "norm_topk_prob": true,
133  "num_attention_heads": 128,
134  "num_experts_per_tok": 8,
135  "num_hidden_layers": 70,
136  "num_key_value_heads": 8,
137  "partial_rotary_factor": 0.334,
138  "rope_theta": 10000000,
139  "routed_scaling_factor": null,
140  "scoring_func": "sigmoid",
141  "sliding_window": 128,
142  "sliding_window_size": 128,
143  "swa_head_dim": 192,
144  "swa_num_attention_heads": 128,
145  "swa_num_key_value_heads": 8,
146  "swa_rope_theta": 10000,
147  "swa_v_head_dim": 128,
148  "tie_word_embeddings": false,
149  "topk_group": 1,
150  "topk_method": "noaux_tc",
151  "torch_dtype": "bfloat16",
152  "transformers_version": "4.57.1",
153  "use_cache": true,
154  "v_head_dim": 128,
155  "vocab_size": 152576
156}