Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,39 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# The Quantized Vicuna 13B v1.5 Model
|
| 6 |
+
|
| 7 |
+
Original Base Model: `lmsys/vicuna-13b-v1.5`.<br>
|
| 8 |
+
Link: [https://huggingface.co/lmsys/vicuna-13b-v1.5](https://huggingface.co/lmsys/vicuna-13b-v1.5)
|
| 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
|
| 39 |
+
Source Codes: [https://github.com/vkola-lab/medpodgpt/tree/main/quantization](https://github.com/vkola-lab/medpodgpt/tree/main/quantization).
|