It's a course playground for "Building Generative AI Applications with Gradio" course from dlai.

Check course website:

https://learn.deeplearning.ai/courses/huggingface-gradio

Models used:

L1: NLP TASK INTERFACE

• dslim/bert-base-NER
• shleifer/distilbart-cnn-12-6

NER(named-entity-recognition) model dslim/bert-base-NER is hosted at Huggingface spaces at following URL: https://huggingface.co/spaces/dolphinium/bert_ner

L2: IMAGE CAPTIONING APP

• Salesforce/blip-image-captioning-base

L3: IMAGE GENERATION APP

• runwayml/stable-diffusion-v1-5

Image generation model stabilityai/stable-diffusion-xl-base-1.0 is hosted at Huggingface spaces at following URL: https://huggingface.co/spaces/dolphinium/stable_diffusion_image_gen

L4: DESCRIBE AND GENERATE GAME

• Salesforce/blip-image-captioning-base
• runwayml/stable-diffusion-v1-5

Describe and Regenerate model TTI STABLE DIFFUSION ITT BLIP-large is hosted at Huggingface spaces at following URL: https://huggingface.co/spaces/dolphinium/Describe_and_Regenerate.

• It is a demonstration of multi model usage on HuggingFace Spaces and Gradio

L5: CHAT WITH ANY LLM

• falcon-40b-instruct (used on course notebook)

Chatbot application mistralai/Mistral-Nemo-Instruct-2407 is hosted at Huggingface spaces at following URL: https://huggingface.co/spaces/dolphinium/mistral_llm_chatbot

since serverless inference api support has been disabled for falcon, I tried these two models:

• mistralai/Mistral-Nemo-Instruct-2407
• meta-llama/Meta-Llama-3.1-8B-Instruct(requires hugging face pro subscription for serverless inference api)