Hello @FrankYoungchen
this is the detailed GPU configuration:
For llama2:
To run the demo, the recommended GPU is A100 with VRAM 40 or higher.
For inference (one video), you can use any P100, P6000, V100 or A100 (Minimum GPU is 16GB)
for Training, you can use A100 with 80GB with batch size=4 or V100 with batch size=1 (Minimum GPU RAM is 32GB)
For Mistral:
To run the demo, the recommended GPU is A100 with VRAM 40 or higher.
For inference (one video), you can use A100, P6000 or V100 (Minimum GPU RAM = 24GB)
for Training, you can only use A100 with 80GB with batch size=1 (Minimum GPU RAM is 80 GB)
For system RAM it is recommended to use RAM=100 GB
Hello @FrankYoungchen
this is the detailed GPU configuration: For llama2: To run the demo, the recommended GPU is A100 with VRAM 40 or higher. For inference (one video), you can use any P100, P6000, V100 or A100 (Minimum GPU is 16GB) for Training, you can use A100 with 80GB with batch size=4 or V100 with batch size=1 (Minimum GPU RAM is 32GB)
For Mistral: To run the demo, the recommended GPU is A100 with VRAM 40 or higher. For inference (one video), you can use A100, P6000 or V100 (Minimum GPU RAM = 24GB) for Training, you can only use A100 with 80GB with batch size=1 (Minimum GPU RAM is 80 GB)
For system RAM it is recommended to use RAM=100 GB