Add files using upload-large-folder tool
Browse files- README.md +1 -79
- tokenizer/tokenizer_config.json +8 -1
README.md
CHANGED
|
@@ -68,85 +68,7 @@ The compression configuration of the model is stored in the `smash_config.json`
|
|
| 68 |
The configuration of the model is stored in the `config.json` file.
|
| 69 |
|
| 70 |
```bash
|
| 71 |
-
{
|
| 72 |
-
"config": {
|
| 73 |
-
"architectures": [
|
| 74 |
-
"Llama4ForCausalLM"
|
| 75 |
-
],
|
| 76 |
-
"attention_bias": false,
|
| 77 |
-
"attention_chunk_size": 8192,
|
| 78 |
-
"attention_dropout": 0.0,
|
| 79 |
-
"attn_scale": 0.1,
|
| 80 |
-
"attn_temperature_tuning": 4,
|
| 81 |
-
"bos_token_id": 200000,
|
| 82 |
-
"cache_implementation": "hybrid",
|
| 83 |
-
"eos_token_id": [
|
| 84 |
-
200001,
|
| 85 |
-
200007,
|
| 86 |
-
200008
|
| 87 |
-
],
|
| 88 |
-
"floor_scale": 8192,
|
| 89 |
-
"for_llm_compressor": false,
|
| 90 |
-
"head_dim": 8,
|
| 91 |
-
"hidden_act": "silu",
|
| 92 |
-
"hidden_size": 16,
|
| 93 |
-
"initializer_range": 0.02,
|
| 94 |
-
"interleave_moe_layer_step": 1,
|
| 95 |
-
"intermediate_size": 32,
|
| 96 |
-
"intermediate_size_mlp": 64,
|
| 97 |
-
"max_position_embeddings": 10485760,
|
| 98 |
-
"model_type": "llama4_text",
|
| 99 |
-
"moe_layers": [
|
| 100 |
-
0,
|
| 101 |
-
1,
|
| 102 |
-
2,
|
| 103 |
-
3,
|
| 104 |
-
4
|
| 105 |
-
],
|
| 106 |
-
"no_rope_layers": [
|
| 107 |
-
1,
|
| 108 |
-
1,
|
| 109 |
-
1,
|
| 110 |
-
0,
|
| 111 |
-
1
|
| 112 |
-
],
|
| 113 |
-
"num_attention_heads": 10,
|
| 114 |
-
"num_experts_per_tok": 1,
|
| 115 |
-
"num_hidden_layers": 5,
|
| 116 |
-
"num_key_value_heads": 2,
|
| 117 |
-
"num_local_experts": 4,
|
| 118 |
-
"output_router_logits": false,
|
| 119 |
-
"pad_token_id": 200018,
|
| 120 |
-
"rms_norm_eps": 1e-05,
|
| 121 |
-
"rope_scaling": {
|
| 122 |
-
"factor": 8.0,
|
| 123 |
-
"high_freq_factor": 4.0,
|
| 124 |
-
"low_freq_factor": 1.0,
|
| 125 |
-
"original_max_position_embeddings": 8192,
|
| 126 |
-
"rope_type": "llama3"
|
| 127 |
-
},
|
| 128 |
-
"rope_theta": 500000.0,
|
| 129 |
-
"router_aux_loss_coef": 0.001,
|
| 130 |
-
"router_jitter_noise": 0.0,
|
| 131 |
-
"tie_word_embeddings": false,
|
| 132 |
-
"torch_dtype": "bfloat16",
|
| 133 |
-
"transformers_version": "4.51.3",
|
| 134 |
-
"use_cache": true,
|
| 135 |
-
"use_qk_norm": true,
|
| 136 |
-
"vocab_size": 202048
|
| 137 |
-
},
|
| 138 |
-
"generation_config": {
|
| 139 |
-
"_from_model_config": true,
|
| 140 |
-
"bos_token_id": 200000,
|
| 141 |
-
"eos_token_id": [
|
| 142 |
-
200001,
|
| 143 |
-
200007,
|
| 144 |
-
200008
|
| 145 |
-
],
|
| 146 |
-
"pad_token_id": 200018,
|
| 147 |
-
"transformers_version": "4.51.3"
|
| 148 |
-
}
|
| 149 |
-
}
|
| 150 |
```
|
| 151 |
|
| 152 |
## 🌍 Join the Pruna AI community!
|
|
|
|
| 68 |
The configuration of the model is stored in the `config.json` file.
|
| 69 |
|
| 70 |
```bash
|
| 71 |
+
{}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 72 |
```
|
| 73 |
|
| 74 |
## 🌍 Join the Pruna AI community!
|
tokenizer/tokenizer_config.json
CHANGED
|
@@ -9086,12 +9086,19 @@
|
|
| 9086 |
"clean_up_tokenization_spaces": false,
|
| 9087 |
"eos_token": "<|eot|>",
|
| 9088 |
"extra_special_tokens": {},
|
|
|
|
| 9089 |
"model_input_names": [
|
| 9090 |
"input_ids",
|
| 9091 |
"attention_mask"
|
| 9092 |
],
|
| 9093 |
"model_max_length": 10485760,
|
|
|
|
| 9094 |
"pad_token": "<|finetune_right_pad_id|>",
|
|
|
|
|
|
|
| 9095 |
"processor_class": "Llama4Processor",
|
| 9096 |
-
"
|
|
|
|
|
|
|
|
|
|
| 9097 |
}
|
|
|
|
| 9086 |
"clean_up_tokenization_spaces": false,
|
| 9087 |
"eos_token": "<|eot|>",
|
| 9088 |
"extra_special_tokens": {},
|
| 9089 |
+
"max_length": 10485760,
|
| 9090 |
"model_input_names": [
|
| 9091 |
"input_ids",
|
| 9092 |
"attention_mask"
|
| 9093 |
],
|
| 9094 |
"model_max_length": 10485760,
|
| 9095 |
+
"pad_to_multiple_of": null,
|
| 9096 |
"pad_token": "<|finetune_right_pad_id|>",
|
| 9097 |
+
"pad_token_type_id": 0,
|
| 9098 |
+
"padding_side": "right",
|
| 9099 |
"processor_class": "Llama4Processor",
|
| 9100 |
+
"stride": 0,
|
| 9101 |
+
"tokenizer_class": "PreTrainedTokenizer",
|
| 9102 |
+
"truncation_side": "right",
|
| 9103 |
+
"truncation_strategy": "longest_first"
|
| 9104 |
}
|