Update README.md
Browse files
README.md
CHANGED
|
@@ -18,4 +18,51 @@ This model is 4bit quantized of glm-4v-9b Model and fixed some error to executin
|
|
| 18 |
|
| 19 |
It has exciting result with less then 10 Giga VRAM (Multi Modal Multi Language).
|
| 20 |
|
| 21 |
-
you can try this model on free google colab. [](https://colab.research.google.com/drive/1aZGX9f5Yw1WbiOrS3TpvPk_UJUP_yYQU?usp=sharing)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
It has exciting result with less then 10 Giga VRAM (Multi Modal Multi Language).
|
| 20 |
|
| 21 |
+
you can try this model on free google colab. [](https://colab.research.google.com/drive/1aZGX9f5Yw1WbiOrS3TpvPk_UJUP_yYQU?usp=sharing)
|
| 22 |
+
|
| 23 |
+
### About GLM-4V-9B
|
| 24 |
+
|
| 25 |
+
GLM-4V-9B is a multimodal language model with visual understanding capabilities. The evaluation results of its related classic tasks are as follows:
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
| | **MMBench-EN-Test** | **MMBench-CN-Test** | **SEEDBench_IMG** | **MMStar** | **MMMU** | **MME** | **HallusionBench** | **AI2D** | **OCRBench** |
|
| 29 |
+
|-------------------------|---------------------|---------------------|-------------------|------------|----------|---------|--------------------|----------|--------------|
|
| 30 |
+
| | 英文综合 | 中文综合 | 综合能力 | 综合能力 | 学科综合 | 感知推理 | 幻觉性 | 图表理解 | 文字识别 |
|
| 31 |
+
| **GPT-4o, 20240513** | 83.4 | 82.1 | 77.1 | 63.9 | 69.2 | 2310.3 | 55 | 84.6 | 736 |
|
| 32 |
+
| **GPT-4v, 20240409** | 81 | 80.2 | 73 | 56 | 61.7 | 2070.2 | 43.9 | 78.6 | 656 |
|
| 33 |
+
| **GPT-4v, 20231106** | 77 | 74.4 | 72.3 | 49.7 | 53.8 | 1771.5 | 46.5 | 75.9 | 516 |
|
| 34 |
+
| **InternVL-Chat-V1.5** | 82.3 | 80.7 | 75.2 | 57.1 | 46.8 | 2189.6 | 47.4 | 80.6 | 720 |
|
| 35 |
+
| **LlaVA-Next-Yi-34B** | 81.1 | 79 | 75.7 | 51.6 | 48.8 | 2050.2 | 34.8 | 78.9 | 574 |
|
| 36 |
+
| **Step-1V** | 80.7 | 79.9 | 70.3 | 50 | 49.9 | 2206.4 | 48.4 | 79.2 | 625 |
|
| 37 |
+
| **MiniCPM-Llama3-V2.5** | 77.6 | 73.8 | 72.3 | 51.8 | 45.8 | 2024.6 | 42.4 | 78.4 | 725 |
|
| 38 |
+
| **Qwen-VL-Max** | 77.6 | 75.7 | 72.7 | 49.5 | 52 | 2281.7 | 41.2 | 75.7 | 684 |
|
| 39 |
+
| **GeminiProVision** | 73.6 | 74.3 | 70.7 | 38.6 | 49 | 2148.9 | 45.7 | 72.9 | 680 |
|
| 40 |
+
| **Claude-3V Opus** | 63.3 | 59.2 | 64 | 45.7 | 54.9 | 1586.8 | 37.8 | 70.6 | 694 |
|
| 41 |
+
| **GLM-4v-9B** | 81.1 | 79.4 | 76.8 | 58.7 | 47.2 | 2163.8 | 46.6 | 81.1 | 786 |
|
| 42 |
+
**This repository is the model repository of GLM-4V-9B, supporting `8K` context length.**
|
| 43 |
+
## Quick Start
|
| 44 |
+
Welcome to visit our [github](https://github.com/THUDM/GLM-4) to view more execution codes.
|
| 45 |
+
```python
|
| 46 |
+
import torch
|
| 47 |
+
from PIL import Image
|
| 48 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 49 |
+
device = "cuda"
|
| 50 |
+
tokenizer = AutoTokenizer.from_pretrained("THUDM/glm-4v-9b", trust_remote_code=True)
|
| 51 |
+
query = 'discribe this image'
|
| 52 |
+
image = Image.open("your image").convert('RGB')
|
| 53 |
+
inputs = tokenizer.apply_chat_template([{"role": "user", "image": image, "content": query}],
|
| 54 |
+
add_generation_prompt=True, tokenize=True, return_tensors="pt",
|
| 55 |
+
return_dict=True) # chat mode
|
| 56 |
+
inputs = inputs.to(device)
|
| 57 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 58 |
+
"THUDM/glm-4v-9b",
|
| 59 |
+
torch_dtype=torch.bfloat16,
|
| 60 |
+
low_cpu_mem_usage=True,
|
| 61 |
+
trust_remote_code=True
|
| 62 |
+
).to(device).eval()
|
| 63 |
+
gen_kwargs = {"max_length": 2500, "do_sample": True, "top_k": 1}
|
| 64 |
+
with torch.no_grad():
|
| 65 |
+
outputs = model.generate(**inputs, **gen_kwargs)
|
| 66 |
+
outputs = outputs[:, inputs['input_ids'].shape[1]:]
|
| 67 |
+
print(tokenizer.decode(outputs[0]))
|
| 68 |
+
```
|