Sagar pallai

sagar007

AI & ML interests

LLM AND STABLE DIFFUSION

Recent Activity

updated a Space 4 days ago
sagar007/Stable_Diffusion_custom
updated a Space 4 days ago
sagar007/Stable_diffusion_Art
liked a Space 5 days ago
sagar007/Multimodal_App
View all activity

Organizations

None yet

sagar007's activity

posted an update 3 months ago
view post
Post
472
Fine-tuned Phi-3.5 Chatbot

This project presents a fine-tuned version of Microsoft's Phi-3.5 model, optimized for enhanced conversational abilities and general knowledge tasks.

Model Details
- Base model: microsoft/Phi-3.5-mini-instruct
- Fine-tuning method: PEFT (Parameter-Efficient Fine-Tuning)
- Training data: [Brief description of your dataset]

Features
- Improved response generation for a wide range of topics
- Enhanced context understanding and coherence
- Optimized for deployment on Hugging Face Spaces

Usage
This model can be used for various natural language processing tasks, including:
- General conversation
- Question answering
- Task instructions
- Creative writing

Try out the model here :
sagar007/phi3.5_mini_instruct_finetune

Limitations
While this fine-tuned model shows improved performance, users should be aware of potential biases and limitations inherent in language models. Always critically evaluate the model's outputs.

Feedback
I welcome any feedback, suggestions, or questions about this project. Feel free to open an issue or contribute to further improvements!

#Phi35 #FineTuning #NLP #MachineLearning #HuggingFace
  • 1 reply
ยท
reacted to their post with ๐Ÿ‘ 3 months ago
view post
Post
1319
Excited to share my new Gradio app featuring the impressive Llama-3.1-Storm-8B model!
This app demonstrates the capabilities of Llama-3.1-Storm-8B, an 8B parameter language model created by Ashvini Kumar Jindal, Pawan Kumar Rajpoot, Ankur Parikh,@akjindal53244
Key highlights of Llama-3.1-Storm-8B:

Outperforms Llama-3.1-8B-Instruct on multiple benchmarks:

Instruction Following (IFEval): +3.93%
Knowledge-driven QA (GPQA): +7.21%
Reduced Hallucinations (TruthfulQA): +9%
Function Calling (BFCL): +7.92%


Achieves impressive results with only 8B parameters
Uses innovative techniques like self-curation and model merging

Try out the model yourself: sagar007/lama_storm_8b

Kudos to the creators for pushing the boundaries of smaller language models! This work makes advanced AI more accessible and efficient.
#AI #NLP #MachineLearning #GradioApp #Llama3
reacted to merve's post with ๐Ÿ‘ 3 months ago
view post
Post
2372
NVIDIA just dropped NVEagle ๐Ÿฆ…

Super impressive vision language model that comes in 7B, 13B and 13B fine-tuned on chat ๐Ÿ’ฌ
Model repositories: merve/nveagle-66d0705108582d73bb235c26
Try it: NVEagle/Eagle-X5-13B-Chat ๐Ÿ’ฌ (works very well! ๐Ÿคฏ)

This model essentially explores having different experts (MoE) for image encoder part of vision language model.
How? ๐Ÿง
The authors concatenate the vision encoder output tokens together, and they apply "pre-alignment" essentially fine-tune experts with frozen text encoder.

Then they freeze both experts and the decoder and just train the projection layer, and finally, they unfreeze everything for supervised fine-tuning โœจ

In the paper, they explore different fusion strategies and vision encoders, extending basic CLIP encoder, and figure out simply concatenating visual tokens works well.
Rest of the architecture is quite similar to LLaVA. (see below the architecture)
posted an update 3 months ago
view post
Post
1319
Excited to share my new Gradio app featuring the impressive Llama-3.1-Storm-8B model!
This app demonstrates the capabilities of Llama-3.1-Storm-8B, an 8B parameter language model created by Ashvini Kumar Jindal, Pawan Kumar Rajpoot, Ankur Parikh,@akjindal53244
Key highlights of Llama-3.1-Storm-8B:

Outperforms Llama-3.1-8B-Instruct on multiple benchmarks:

Instruction Following (IFEval): +3.93%
Knowledge-driven QA (GPQA): +7.21%
Reduced Hallucinations (TruthfulQA): +9%
Function Calling (BFCL): +7.92%


Achieves impressive results with only 8B parameters
Uses innovative techniques like self-curation and model merging

Try out the model yourself: sagar007/lama_storm_8b

Kudos to the creators for pushing the boundaries of smaller language models! This work makes advanced AI more accessible and efficient.
#AI #NLP #MachineLearning #GradioApp #Llama3
replied to their post 3 months ago
posted an update 3 months ago
view post
Post
613
๐Ÿ“ฃ New Project Alert: Phi 3.5 Multimodal AI Demo ๐ŸŽ‰
Excited to share my latest project that combines the power of Phi 3.5 text and vision models with text-to-speech capabilities!
๐Ÿ”‘ Key Features:
1๏ธโƒฃ Phi 3.5 Text Model for dynamic conversations
2๏ธโƒฃ Phi 3.5 Vision Model for advanced image analysis
3๏ธโƒฃ Text-to-Speech integration for an audio dimension
๐Ÿ› ๏ธ Tech Stack:

Transformers
Gradio
PyTorch
Flash Attention 2
Parler TTS

This project demonstrates the potential of integrating multiple AI models to create a more comprehensive and interactive user experience. It's a step towards more natural and versatile AI assistants.
๐Ÿ‘‰ Check out the demo and let me know your thoughts! How would you extend this project?
๐Ÿ”— Demo Link: sagar007/Multimodal_App
#MultimodalAI #PhiModel #MachineLearning #AIDemo
  • 1 reply
ยท
posted an update 4 months ago
view post
Post
713
๐Ÿ“ AI Math Equation Solver: Your Step-by-Step Solution Companion

Hello Hugging Face community! ๐Ÿ‘‹ I'm excited to share my latest Space: the AI Math Equation Solver!

๐Ÿ” What does it do?

This Space uses the power of AI to solve math problems from images. Simply upload a picture of a math equation or problem, and the AI will provide a detailed, step-by-step solution. It's perfect for students, teachers, or anyone looking to understand complex mathematical concepts better.

๐Ÿง  How does it work?

- Backend: Utilizes the microsoft/Phi-3.5-vision-instruct model for image understanding and mathematical reasoning.
- Frontend: Built with Gradio for a clean, user-friendly interface.
- Features:
- Image upload for math problems
- Detailed step-by-step solutions
- Example problems to try instantly

๐Ÿš€ Try it out! sagar007/phi-vision-math-assistant

Visit the Space here: [Insert your Hugging Face Space URL]

๐Ÿ’ก Use cases:

- Students: Check your work or get help with homework
- Teachers: Create detailed solution guides quickly
- Tutors: Explain complex problems more effectively
- Self-learners: Understand new mathematical concepts

๐Ÿ› ๏ธ Technical Details:

- Model: microsoft/Phi-3.5-vision-instruct
- Libraries: transformers, Gradio, PyTorch
- Optimizations: Uses Flash Attention for improved performance

๐Ÿค Contribute:

This is an open project, and I welcome contributions! Whether it's improving the model, enhancing the UI, or adding new features, feel free to fork the project and submit your pull requests.

๐Ÿ“ฃ Feedback:

I'd love to hear your thoughts! How are you using this Space? Any suggestions for improvements? Let me know in the comments below.

Happy problem-solving! ๐ŸŽ‰

#MachineLearning #AI #Mathematics #Education #HuggingFace
posted an update 4 months ago
view post
Post
750
๐ŸŒŸ Enchanted Tales Generator: A GPT-2 Inspired Story Weaver ๐ŸŒŸ

Hello Hugging Face community! I'm excited to share my latest project, the Enchanted Tales Generator, inspired by Andrej Karpathy's enlightening YouTube videos on GPT-2.

๐Ÿง™โ€โ™‚๏ธ What is the Enchanted Tales Generator?

The Enchanted Tales Generator is a magical text generation app that weaves whimsical stories based on your prompts. It's powered by a fine-tuned GPT model and brings the wonder of AI-generated storytelling to your fingertips.

๐ŸŽฅ Inspiration

This project was born from the inspiration I drew from Andrej Karpathy's incredible YouTube series on GPT-2. His clear explanations and deep insights into the workings of language models sparked my imagination and drove me to create something that could bring joy and creativity to others.

๐Ÿ”ฎ Features

- Story Incantation: Input your magical prompt to start your tale
- Scroll Length: Adjust the length of your generated story
- Magical Intensity: Control the creativity (temperature) of the generation
- Arcane Diversity: Fine-tune the variety of word choices (top-k sampling)

๐Ÿš€ Try It Out: sagar007/GPT-2-with_gpu

You can experience the Enchanted Tales Generator right here on Hugging Face Spaces: [Insert your Spaces link here]

๐Ÿ› ๏ธ Behind the Scenes

The app is built using:
- PyTorch for the underlying GPT model
- Gradio for the user interface
- Hugging Face Spaces for deployment

I've implemented techniques learned from Karpathy's videos, such as:
- Custom GPT architecture
- Top-k sampling for diverse outputs
- Temperature-controlled generation

๐ŸŒˆ Future Enchantments

I'm continuously working on improving the Enchanted Tales Generator. Some ideas for future updates include:
- Fine-tuning on specific genres of stories
- Adding options for character and setting generation
- Implementing more advanced sampling techniques

๐Ÿค Join the Magic!