-
CODE:
```python
from PIL import Image
import torch
import os
from llava.serve.classes.Utils import *
from llava.serve.classes.Compiler import *
from llava.model.builder import load_mixed_…
-
### Describe the issue
Issue:
We are trying to finetune the model on our dataset.
Currently, we are able to successfully finetune model `lmsys/vicuna-13b-v1.5` using projector weights `llava-v…
-
# How would you like the feature to work?
Add support for projectors in the PrimitiveParams family of scripting functions.
[ PRIM_PROJECTOR, string texture, float fov, float focus, float ambience …
-
### Question
when I run v1.5/finetune_lora.sh,it raise error about mm projecoter:
According t…
-
### Question
Hi! Can you please tell how many visual tokens are given to LM? Is it 576? If it's something else can you please show it in code where the output of clip is transformed except the projec…
-
### Question
using script scripts/v1_5/pretrain.sh, I conducted llava pretraining using some task-specific image-text pair, now I am thinking of testing the pretrained model, and wondering how can …
-
Hi @tsb0601 @s9xie , thanks for your great projector.
I would like to inquire about the process of reproducing the result of Figure 6 (i.e., MLLM’s performance on visual patterns). In scripts/evalu…
-
### Quick summary
The third installment of the Ratchet & Clank Trilogy keeps crashing after playing for a few minutes.
### Details
I recently tried to pick up Ratchet & Clank 3 again to continue wh…
-
### Describe the issue
Issue: I try to do visual instruction tuning using the pretrained projector liuhaotian/llava-pretrain-llama-2-7b-chat. However, got the following issue. I have download the pr…
llv22 updated
5 months ago
-
If I click the projector button, it toggles for all open windows, as opposed to just the one that I enabled it on.
It seems like the functionality only does it for that window still, but the visual f…