Update README.md
Browse files
README.md
CHANGED
|
@@ -9,7 +9,30 @@ Link: [https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct](https://huggingf
|
|
| 9 |
|
| 10 |
## Quantization Configurations
|
| 11 |
```
|
| 12 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
```
|
| 14 |
|
| 15 |
## Source Codes
|
|
|
|
| 9 |
|
| 10 |
## Quantization Configurations
|
| 11 |
```
|
| 12 |
+
"quantization_config": {
|
| 13 |
+
"batch_size": 1,
|
| 14 |
+
"bits": 4,
|
| 15 |
+
"block_name_to_quantize": null,
|
| 16 |
+
"cache_block_outputs": true,
|
| 17 |
+
"damp_percent": 0.1,
|
| 18 |
+
"dataset": null,
|
| 19 |
+
"desc_act": false,
|
| 20 |
+
"exllama_config": {
|
| 21 |
+
"version": 1
|
| 22 |
+
},
|
| 23 |
+
"group_size": 128,
|
| 24 |
+
"max_input_length": null,
|
| 25 |
+
"model_seqlen": null,
|
| 26 |
+
"module_name_preceding_first_block": null,
|
| 27 |
+
"modules_in_block_to_quantize": null,
|
| 28 |
+
"pad_token_id": null,
|
| 29 |
+
"quant_method": "gptq",
|
| 30 |
+
"sym": true,
|
| 31 |
+
"tokenizer": null,
|
| 32 |
+
"true_sequential": true,
|
| 33 |
+
"use_cuda_fp16": false,
|
| 34 |
+
"use_exllama": true
|
| 35 |
+
},
|
| 36 |
```
|
| 37 |
|
| 38 |
## Source Codes
|