wyclike

could it finetune in qlora or quant 4/8?

1

i have only a A40(48GB),how could i do to finetune the model? Althought i use lora ,i still meet the problem that "cuda out of memory". So i want to...

https://huggingface.co/runwayml/stable-diffusion-v1-5 404

2

请问谁有保留这个ckpt吗，huggingface上这个模型权重被撤了

About model ckpt

1

could you please provide the post-supervised checkpoint for the llama-7b and flan-t5-xxl?

视频token占用数疑问

2

当我把fps设为1.0，video_max_pixels和video_min_pixels设置为336*336（即每帧固定为336*336的分辨率），然后把video_min_frames和video_max_frames都设为400（即把一个长视频卡死到400帧，原视频远超400帧），文本token数大概为10左右。然后我计算总token数，在这个情况下，理论上单帧token：(336*336)/(28*28)=144，这个视频应该为400*144=57600,但实测出来大概为28800左右。 请问是我对于qwen2vl的架构理解不清晰嘛？

After insert special tokens，Have changed lm_head dimension？

1

After reading the code of `finetune.py`, I am confused about the fact that after adding special tokens, only the embedding dimensions were modified, but it seems that the dimensions of...

请问可以提供'../output/flant5-11b-v8-zeroshot/predict_eval_predictions.jsonl'文件嘛？我想看看数据格式并跑通caculate_f1.py

3

如题，我主要想看看如何评测