tsinghua-ee

video-SALMONN-2_plus_72B

README

License: apache-2.0

Results

How to Use

IMPORTANT: To get the same evaluation result, please use --max_frames 768 --max_pixels 61250. Using excessively high resolution or frame rate for evaluation may lead to too much input token count for the model, potentially causing performance degradation.

Prepare the dataset following scripts/example_av.json, scripts/example_v.json, scripts/example_dpo.json, and scripts/example_a.json
Prepare base audio model through modifying the path in gen_audio_model.py

To conduct audio alignment, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name audio_alignment --dataset path_to_dataset --lr 2e-5 --train_qformer --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --bs 16 --epoch 5 --save_steps 5000

To conduct audio-visual SFT, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name av_sft --dataset path_to_dataset --lr 2e-5 --train_qformer --train_proj --max_frames 768 --max_pixels 61250 --model audio_align_model --model_base path_to_audio_model --epoch 5 --save_steps 2000 --use_lora --lora_r 128 --lora_alpha 256

To conduct DPO, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name dpo --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model audio_visual_base --model_base audio_align_model --lora_ckpt audio_visual_checkpoint --train_type gdpo --use_lora --lora_r 128 --lora_alpha 256 --lr 5e-6 --epoch 1 --save_steps 200 --train_qformer --train_proj

To evaluate 7B model, use the following script:

bash

bash scripts/test.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt

To evaluate 72B model, use the following script:

bash

bash scripts/test_8.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

tsinghua-ee

Model Tree

Base

Qwen/Qwen2.5-VL-72B-Instruct

Adapter

this model

Input Modalities

TextImage

Output Modalities

Text

Supported Functionality

Dedicated Endpoints

Explore FriendliAI today

Get started Talk to an engineer

README

License: apache-2.0

Results

How to Use

Prepare the dataset following scripts/example_av.json, scripts/example_v.json, scripts/example_dpo.json, and scripts/example_a.json
Prepare base audio model through modifying the path in gen_audio_model.py

To conduct audio alignment, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name audio_alignment --dataset path_to_dataset --lr 2e-5 --train_qformer --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --bs 16 --epoch 5 --save_steps 5000

bash scripts/train.sh --interval 0.1 --run_name audio_alignment --dataset path_to_dataset --lr 2e-5 --train_qformer --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --bs 16 --epoch 5 --save_steps 5000

To conduct audio-visual SFT, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name av_sft --dataset path_to_dataset --lr 2e-5 --train_qformer --train_proj --max_frames 768 --max_pixels 61250 --model audio_align_model --model_base path_to_audio_model --epoch 5 --save_steps 2000 --use_lora --lora_r 128 --lora_alpha 256

bash scripts/train.sh --interval 0.1 --run_name av_sft --dataset path_to_dataset --lr 2e-5 --train_qformer --train_proj --max_frames 768 --max_pixels 61250 --model audio_align_model --model_base path_to_audio_model --epoch 5 --save_steps 2000 --use_lora --lora_r 128 --lora_alpha 256

To conduct DPO, use the following script:

bash

bash scripts/train.sh --interval 0.1 --run_name dpo --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model audio_visual_base --model_base audio_align_model --lora_ckpt audio_visual_checkpoint --train_type gdpo --use_lora --lora_r 128 --lora_alpha 256 --lr 5e-6 --epoch 1 --save_steps 200 --train_qformer --train_proj

bash scripts/train.sh --interval 0.1 --run_name dpo --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model audio_visual_base --model_base audio_align_model --lora_ckpt audio_visual_checkpoint --train_type gdpo --use_lora --lora_r 128 --lora_alpha 256 --lr 5e-6 --epoch 1 --save_steps 200 --train_qformer --train_proj

To evaluate 7B model, use the following script:

bash

bash scripts/test.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt

bash scripts/test.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt

To evaluate 72B model, use the following script:

bash

bash scripts/test_8.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt

bash scripts/test_8.sh --interval 0.1 --run_name eval --dataset path_to_dataset --max_frames 768 --max_pixels 61250 --model path_to_audio_model --model_base path_to_audio_model --lora_ckpt model_ckpt