Update requirements.txt

try to load gemma again

testing fps for internvl3 processor

added do_sample to generate

make sure temperature is float

skipping the use of gemma model for now

added temp and testing gemma

updated case matching

testing model quantization

skip loading of internvl3 8b model

adding more internvl3 models

debugging internvl3

debugging internvl3

debugging internvl3

fixed push logic in the last step

updated to not generate requirements if it already exists

added different inference code for internvl3

added intern video model

added control for fps and max tokens

updated app to load multiple models

Update requirements.txt

Update requirements.txt

pinning transformer to try to fix flash attention error

added in extra requirements

switch back to previous app

attempt to slim down req

Update requirements.txt

Update requirements.txt

chore: update requirements.txt [auto-generated by CI]

updated torch dependency

trying newer torch versions

trying a different inference script

trying a different inference script

try quantization again

try quantization again

trying to load model and processors outside space decorator

trying model quantization

trying model quantization

fixing issue with device map for the inputs

added low_cpu_mem_usage and move input to device also

added low_cpu_mem_usage and move input to device also

testing more efficient model loading

testing more efficient model loading

pinning transformers version

make sure DTYPE is used

testing cuda for processor

added back in input.to(device)

make flash attention an input

try use flash attention again

trying to set device outsite of spaces.GPU decorator