Waseem AlShikh's picture

Waseem AlShikh

wassemgtk

·

https://writer.com/

AI & ML interests

Multi-modal, Palmyra LLMs, Knowledge Graph

Recent Activity

replied to their post 9 days ago

# GESAL: Real-Time Adaptation for LLMs We’re excited to unveil **Graph-Enhanced Singular Adaptive Learning (GESAL)**, a framework that lets LLMs like `meta-llama/Llama-3.2-1B` adapt in real time using user feedback. Check out the code and white paper on GitHub! 🔗 **Code**: [https://github.com/writer/AI-Adaptive-Learning-GESAL](https://github.com/writer/AI-Adaptive-Learning-GESAL) --- ## Why GESAL? Static LLMs struggle to adapt without heavy retraining. GESAL solves this with: - **SVF**: Adapts weights via \( W' = U (\Sigma \cdot z) V^T \), using few parameters. - **Graph Memory**: Stores adaptations in nodes for scalability. - **RL**: Updates via \( J(z) = \mathbb{E}[\log \pi_z(y|x) r] \) based on feedback. --- ## How It Works Ask "How many R’s in ‘strawberry’?" If it says "2" and you say "no," GESAL learns to say "3" next time, avoiding repeats. --- ## Try It Built with Hugging Face’s `transformers`: ```bash pip install transformers torch numpy python Adaptive_Learning_(GESAL).py ``` Needs a Hugging Face token for Llama-3.2-1B. --- ## Results GESAL hits 95% accuracy after 5 feedbacks vs. LoRA’s 70%. It’s efficient (~0.5M params) and scalable.

replied to their post 9 days ago

# GESAL: Real-Time Adaptation for LLMs We’re excited to unveil **Graph-Enhanced Singular Adaptive Learning (GESAL)**, a framework that lets LLMs like `meta-llama/Llama-3.2-1B` adapt in real time using user feedback. Check out the code and white paper on GitHub! 🔗 **Code**: [https://github.com/writer/AI-Adaptive-Learning-GESAL](https://github.com/writer/AI-Adaptive-Learning-GESAL) --- ## Why GESAL? Static LLMs struggle to adapt without heavy retraining. GESAL solves this with: - **SVF**: Adapts weights via \( W' = U (\Sigma \cdot z) V^T \), using few parameters. - **Graph Memory**: Stores adaptations in nodes for scalability. - **RL**: Updates via \( J(z) = \mathbb{E}[\log \pi_z(y|x) r] \) based on feedback. --- ## How It Works Ask "How many R’s in ‘strawberry’?" If it says "2" and you say "no," GESAL learns to say "3" next time, avoiding repeats. --- ## Try It Built with Hugging Face’s `transformers`: ```bash pip install transformers torch numpy python Adaptive_Learning_(GESAL).py ``` Needs a Hugging Face token for Llama-3.2-1B. --- ## Results GESAL hits 95% accuracy after 5 feedbacks vs. LoRA’s 70%. It’s efficient (~0.5M params) and scalable.

replied to their post 9 days ago

# GESAL: Real-Time Adaptation for LLMs We’re excited to unveil **Graph-Enhanced Singular Adaptive Learning (GESAL)**, a framework that lets LLMs like `meta-llama/Llama-3.2-1B` adapt in real time using user feedback. Check out the code and white paper on GitHub! 🔗 **Code**: [https://github.com/writer/AI-Adaptive-Learning-GESAL](https://github.com/writer/AI-Adaptive-Learning-GESAL) --- ## Why GESAL? Static LLMs struggle to adapt without heavy retraining. GESAL solves this with: - **SVF**: Adapts weights via \( W' = U (\Sigma \cdot z) V^T \), using few parameters. - **Graph Memory**: Stores adaptations in nodes for scalability. - **RL**: Updates via \( J(z) = \mathbb{E}[\log \pi_z(y|x) r] \) based on feedback. --- ## How It Works Ask "How many R’s in ‘strawberry’?" If it says "2" and you say "no," GESAL learns to say "3" next time, avoiding repeats. --- ## Try It Built with Hugging Face’s `transformers`: ```bash pip install transformers torch numpy python Adaptive_Learning_(GESAL).py ``` Needs a Hugging Face token for Llama-3.2-1B. --- ## Results GESAL hits 95% accuracy after 5 feedbacks vs. LoRA’s 70%. It’s efficient (~0.5M params) and scalable.

View all activity

Organizations

wassemgtk's activity

New activity in Writer/Financial_LLM_Performance_Leaderboard 19 days ago

Added explanation of how combined score is calculated

#2 opened 19 days ago by

Update app.py

#1 opened 19 days ago by

New activity in Writer/FailSafeQA 25 days ago

Add link to paper and update citation.

#2 opened 25 days ago by

New activity in Writer/Palmyra-Creative 4 months ago

Any details on this and the base model?

#1 opened 4 months ago by

New activity in fireworks-ai/llama-3-firefunction-v2 5 months ago

Llama3.1

#4 opened 5 months ago by

New activity in Writer/Palmyra-Med-70B-32K 6 months ago

Base model?

#4 opened 7 months ago by

New activity in philschmid/llm-pricing 6 months ago

Update src/lib/data.ts

#11 opened 6 months ago by

New activity in Writer/Palmyra-Med-70B-32K 7 months ago

Access the model via Ollama

#3 opened 7 months ago by

Quantization allowed?

#1 opened 7 months ago by

New activity in Writer/palmyra-3B 7 months ago

Adding `safetensors` variant of this model

#1 opened about 1 year ago by

New activity in philschmid/llm-pricing 8 months ago

Update src/lib/data.ts

#2 opened 8 months ago by

Adding Writer LLMs

#1 opened 8 months ago by

New activity in openlifescienceai/open_medical_llm_leaderboard 10 months ago

70B submissions are just disappearing after a while

#10 opened 10 months ago by

It's broken

#9 opened 10 months ago by

New activity in colbert-ir/colbertv2.0 over 1 year ago

Create README.md

#2 opened over 1 year ago by

New activity in Writer/InstructPalmyra-20b over 1 year ago

Create handler.py

#4 opened over 1 year ago by

New activity in Writer/palmyra-med-20b over 1 year ago

add example for text-generation-inference

#1 opened over 1 year ago by

New activity in Writer/palmyra-large over 1 year ago

add text-generation-inference example

#1 opened over 1 year ago by

New activity in Writer/InstructPalmyra-20b over 1 year ago

add example for text-generation-inference

#3 opened over 1 year ago by

changed palmyra-med -> InstructPalmyra-20b

#2 opened over 1 year ago by