Add files using upload-large-folder tool
Browse files
README.md
CHANGED
@@ -53,92 +53,6 @@ The compression configuration of the model is stored in the `smash_config.json`
|
|
53 |
}
|
54 |
```
|
55 |
|
56 |
-
## Model Configuration
|
57 |
-
|
58 |
-
The configuration of the model is stored in the `config.json` file.
|
59 |
-
|
60 |
-
```bash
|
61 |
-
{
|
62 |
-
"config": {
|
63 |
-
"architectures": [
|
64 |
-
"Llama4ForCausalLM"
|
65 |
-
],
|
66 |
-
"attention_bias": false,
|
67 |
-
"attention_chunk_size": 8192,
|
68 |
-
"attention_dropout": 0.0,
|
69 |
-
"attn_scale": 0.1,
|
70 |
-
"attn_temperature_tuning": 4,
|
71 |
-
"bos_token_id": 200000,
|
72 |
-
"cache_implementation": "hybrid",
|
73 |
-
"eos_token_id": [
|
74 |
-
200001,
|
75 |
-
200007,
|
76 |
-
200008
|
77 |
-
],
|
78 |
-
"floor_scale": 8192,
|
79 |
-
"for_llm_compressor": false,
|
80 |
-
"head_dim": 8,
|
81 |
-
"hidden_act": "silu",
|
82 |
-
"hidden_size": 16,
|
83 |
-
"initializer_range": 0.02,
|
84 |
-
"interleave_moe_layer_step": 1,
|
85 |
-
"intermediate_size": 32,
|
86 |
-
"intermediate_size_mlp": 64,
|
87 |
-
"max_position_embeddings": 10485760,
|
88 |
-
"model_type": "llama4_text",
|
89 |
-
"moe_layers": [
|
90 |
-
0,
|
91 |
-
1,
|
92 |
-
2,
|
93 |
-
3,
|
94 |
-
4
|
95 |
-
],
|
96 |
-
"no_rope_layers": [
|
97 |
-
1,
|
98 |
-
1,
|
99 |
-
1,
|
100 |
-
0,
|
101 |
-
1
|
102 |
-
],
|
103 |
-
"num_attention_heads": 10,
|
104 |
-
"num_experts_per_tok": 1,
|
105 |
-
"num_hidden_layers": 5,
|
106 |
-
"num_key_value_heads": 2,
|
107 |
-
"num_local_experts": 4,
|
108 |
-
"output_router_logits": false,
|
109 |
-
"pad_token_id": 200018,
|
110 |
-
"rms_norm_eps": 1e-05,
|
111 |
-
"rope_scaling": {
|
112 |
-
"factor": 8.0,
|
113 |
-
"high_freq_factor": 4.0,
|
114 |
-
"low_freq_factor": 1.0,
|
115 |
-
"original_max_position_embeddings": 8192,
|
116 |
-
"rope_type": "llama3"
|
117 |
-
},
|
118 |
-
"rope_theta": 500000.0,
|
119 |
-
"router_aux_loss_coef": 0.001,
|
120 |
-
"router_jitter_noise": 0.0,
|
121 |
-
"tie_word_embeddings": false,
|
122 |
-
"torch_dtype": "float32",
|
123 |
-
"transformers_version": "4.51.3",
|
124 |
-
"use_cache": true,
|
125 |
-
"use_qk_norm": true,
|
126 |
-
"vocab_size": 202048
|
127 |
-
},
|
128 |
-
"generation_config": {
|
129 |
-
"_from_model_config": true,
|
130 |
-
"bos_token_id": 200000,
|
131 |
-
"eos_token_id": [
|
132 |
-
200001,
|
133 |
-
200007,
|
134 |
-
200008
|
135 |
-
],
|
136 |
-
"pad_token_id": 200018,
|
137 |
-
"transformers_version": "4.51.3"
|
138 |
-
}
|
139 |
-
}
|
140 |
-
```
|
141 |
-
|
142 |
## 🌍 Join the Pruna AI community!
|
143 |
|
144 |
[](https://twitter.com/PrunaAI)
|
|
|
53 |
}
|
54 |
```
|
55 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
## 🌍 Join the Pruna AI community!
|
57 |
|
58 |
[](https://twitter.com/PrunaAI)
|