InternVL

InternVL copied to clipboard

請問模型輸出是否有提供 confidence score 的參數？

2

### Motivation 我是使用 hugging face 方式推論 OpenGVLab/InternVL2-8B 模型，請問模型輸出是否有提供 confidence score 的參數？ 謝謝！ ### Related resources _No response_ ### Additional context _No response_

st931630

### 📚 The doc issue 请问InternVL2的grounding和多图任务使用了哪些数据集？预训练和微调阶段都涵盖了这两类任务吗？ ### Suggest a potential alternative/fix _No response_

QingrongH

[Feature] 使用internvl_chat_llava/scripts_internvl/pretrain_internvit6b_448_vicuna7b.sh预训练时train/grad_norm值0.5左右loss在0.5与2.4之间振荡

1

### Motivation 如题，grad_norm很快降为0.5左右，是什么参数不对吗？ 资源：1*A100 80G 训练参数： deepspeed --include localhost:2 \ llava/train/train_mem.py \ --deepspeed ./scripts/zero2.json \ --model_name_or_path $DATA_HOME/pretrained_mm_projector/vicuna-7b-v1.5 \ --version plain \ --data_path $DATA_HOME/LLaVA-Pretrain/enhanced_llava_pretrain_data_708K.json \ --image_folder $DATA_HOME/LLaVA-Pretrain/images \ --vision_tower $DATA_HOME/pretrained_mm_projector/InternViT-300M-448px \...

humphery755

[Docs] image-text retrieval

1

### 📚 The doc issue I'm confused about the internvl_c_retrieval and internvl_g_retrieval ,what's the different? and the video_retrieval can an example replace it? ### Suggest a potential alternative/fix could you...

azuredsky

[Docs] 关于输出长度的prompt

1

### 📚 The doc issue 请问如果想让模型输出固定长度（大概在这个范围即可），比如输出长度为200个token，该怎么给prompt比较好 ### Suggest a potential alternative/fix 需要解答

Jse-NGV

### Motivation 您好， 我想返回 InternVL2‑26B的output_attentions，但是使用transformer推理时，发现，在模型自带的代码里面，不论外面设置output_attentions为True或者False，modeling_internlm2.py代码里面都将其人工设定为False，导致无法返回attn_weights，如果要返回有效的attn_weights，应该怎么修改代码？ 感谢。 ### Related resources _No response_ ### Additional context _No response_

zhly0

I have made some experiments on the chat demo, and found InternVL2-Pro is amazing! Could you tell how large InternVL2-Pro is and when to release its model? And will there...

cxmscb

KeyError: 'architectures'

2

### Checklist - [ ] 1. I have searched related issues but cannot get the expected help. - [ ] 2. The bug has not been fixed in the latest...

CachCheng

### Checklist - [X] 1. I have searched related issues but cannot get the expected help. - [X] 2. The bug has not been fixed in the latest version. -...

tcexeexe

直接用device_map="auto" 或 device_map="balanced_low_0" 会报错，text_embeds = text_embeds[torch.arange(text_embeds.shape[0]), attention_mask.sum(1) - 1] RuntimeError: indices should be either on cpu or on the same device as the indexed tensor (cuda:1)

lisenjie757