Update README.md
Browse files
README.md
CHANGED
@@ -87,6 +87,18 @@ content = "这张图片代表哪家公司?"
|
|
87 |
## The image represents the company Intel.
|
88 |
```
|
89 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
90 |
|
91 |
|
92 |
### Generate the model
|
|
|
87 |
## The image represents the company Intel.
|
88 |
```
|
89 |
|
90 |
+
## Evaluation the model
|
91 |
+
pip3 install lmms_eval. The evaluation process may encounter errors that require changing model backend or evaluation code. Detailed instructions will be provided in a future update
|
92 |
+
```bash
|
93 |
+
auto-round-mllm --lmms --model OPEA/cogvlm2-llama3-chat-19B-qvision-int4-sym-inc --tasks pope,textvqa_val,scienceqa,mmbench_en --output_dir "./eval_result" --device cuda:0
|
94 |
+
```
|
95 |
+
|Metric |16bits| Llava Calib INT4 |
|
96 |
+
|:-------------------|:------|:--------------|
|
97 |
+
|avg |80.38 |80.37 |
|
98 |
+
|MMBench_DEV_EN_V11 |75.86 |74.57 |
|
99 |
+
|ScienceQA_VAL |-|- |
|
100 |
+
|TextVQA_VAL |77.77 |78.91 |
|
101 |
+
|POPE |87.37 |87.64 |
|
102 |
|
103 |
|
104 |
### Generate the model
|