An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Sorted by Tags
AI
Agents
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Let Agents do the talking: A Scalable Way to Evaluate Multi-Turn Chatbots
Interactive evaluations: lightweight, automated tests that use agents to measure multi-turn chatbot quality at scale.
Annotation
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Async
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
CUDA
CUDA Study Log 4: Optimizing Constrained Decoding with Triton Kernel
Update traditional CUDA matrix multiplication kernel for constrained decoding
CUDA Studylog 3 - Tiling and Shared Memory for Matrix Multiplication Optimization
Optimizing CUDA matrix multiplication using tiling and shared memory, with detailed explanations of memory access patterns and performance improvements
CUDA Studylog 2 - Matrix Multiplication and 2D Grid Organization
Deep dive into implementing efficient matrix multiplication using CUDA, with a focus on memory optimization techniques
CUDA Studylog 1 - Getting a taste of CUDA kernels
A Introduction Guide for ML Engineers. Learn the fundamentals and practical implementations needed to get started with CUDA kernels
[Shorts-3] GPU Memory management -> Automatically select GPU with unused RAM
Python code snippet to automatically select the GPU out of a GPU-pool
Chatbots
Let Agents do the talking: A Scalable Way to Evaluate Multi-Turn Chatbots
Interactive evaluations: lightweight, automated tests that use agents to measure multi-turn chatbot quality at scale.
Concurrency
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
Constrained Decoding
CUDA Study Log 4: Optimizing Constrained Decoding with Triton Kernel
Update traditional CUDA matrix multiplication kernel for constrained decoding
Coroutines
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
DVC
Version Controlling ML projects using DVC
Automate ML steps using pipelines
Data Science
Reducing the runtime from 2+ hours (on-GPU) to 8 mins in a recent project
Highlighting the bad engineering practices in production ready ML code
Version Controlling ML projects using DVC
Automate ML steps using pipelines
Data Versioning
Version Controlling ML projects using DVC
Automate ML steps using pipelines
Data-Augmentation
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
Data-Science
[Shorts-3] GPU Memory management -> Automatically select GPU with unused RAM
Python code snippet to automatically select the GPU out of a GPU-pool
Dataset creation
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Deep Learning
[Shorts-1] How to download HuggingFace models the right way
Downloading huggingface models as git repos
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Engineering
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
Environments
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Evaluation
Let Agents do the talking: A Scalable Way to Evaluate Multi-Turn Chatbots
Interactive evaluations: lightweight, automated tests that use agents to measure multi-turn chatbot quality at scale.
Event Loop
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
GPT
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
GPT3
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
GPT4
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
GPU
CUDA Studylog 3 - Tiling and Shared Memory for Matrix Multiplication Optimization
Optimizing CUDA matrix multiplication using tiling and shared memory, with detailed explanations of memory access patterns and performance improvements
CUDA Studylog 2 - Matrix Multiplication and 2D Grid Organization
Deep dive into implementing efficient matrix multiplication using CUDA, with a focus on memory optimization techniques
CUDA Studylog 1 - Getting a taste of CUDA kernels
A Introduction Guide for ML Engineers. Learn the fundamentals and practical implementations needed to get started with CUDA kernels
[Shorts-3] GPU Memory management -> Automatically select GPU with unused RAM
Python code snippet to automatically select the GPU out of a GPU-pool
GRPO
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
HuggingFace
[Shorts-1] How to download HuggingFace models the right way
Downloading huggingface models as git repos
ICV
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
In-Context Vectors
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
Inference
Reducing the runtime from 2+ hours (on-GPU) to 8 mins in a recent project
Highlighting the bad engineering practices in production ready ML code
Jupyter Notebook
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
LLM
Let Agents do the talking: A Scalable Way to Evaluate Multi-Turn Chatbots
Interactive evaluations: lightweight, automated tests that use agents to measure multi-turn chatbot quality at scale.
CUDA Study Log 4: Optimizing Constrained Decoding with Triton Kernel
Update traditional CUDA matrix multiplication kernel for constrained decoding
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
LMs
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
Language Models
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
LoRA
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
ML
CUDA Studylog 3 - Tiling and Shared Memory for Matrix Multiplication Optimization
Optimizing CUDA matrix multiplication using tiling and shared memory, with detailed explanations of memory access patterns and performance improvements
CUDA Studylog 2 - Matrix Multiplication and 2D Grid Organization
Deep dive into implementing efficient matrix multiplication using CUDA, with a focus on memory optimization techniques
CUDA Studylog 1 - Getting a taste of CUDA kernels
A Introduction Guide for ML Engineers. Learn the fundamentals and practical implementations needed to get started with CUDA kernels
Machine Learning
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Model Serialization
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
Model management
Version Controlling ML projects using DVC
Automate ML steps using pipelines
NLP
[Shorts-1] How to download HuggingFace models the right way
Downloading huggingface models as git repos
Adapting Text Augmentation to Industry problems
Applying recent NLP-Augmentation ideas to Industrial projects
Notebook
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
OpenAI
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Optimization
CUDA Study Log 4: Optimizing Constrained Decoding with Triton Kernel
Update traditional CUDA matrix multiplication kernel for constrained decoding
CUDA Studylog 3 - Tiling and Shared Memory for Matrix Multiplication Optimization
Optimizing CUDA matrix multiplication using tiling and shared memory, with detailed explanations of memory access patterns and performance improvements
CUDA Studylog 2 - Matrix Multiplication and 2D Grid Organization
Deep dive into implementing efficient matrix multiplication using CUDA, with a focus on memory optimization techniques
CUDA Studylog 1 - Getting a taste of CUDA kernels
A Introduction Guide for ML Engineers. Learn the fundamentals and practical implementations needed to get started with CUDA kernels
Reducing the runtime from 2+ hours (on-GPU) to 8 mins in a recent project
Highlighting the bad engineering practices in production ready ML code
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
PDF parsing
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
Performance
CUDA Studylog 3 - Tiling and Shared Memory for Matrix Multiplication Optimization
Optimizing CUDA matrix multiplication using tiling and shared memory, with detailed explanations of memory access patterns and performance improvements
CUDA Studylog 2 - Matrix Multiplication and 2D Grid Organization
Deep dive into implementing efficient matrix multiplication using CUDA, with a focus on memory optimization techniques
CUDA Studylog 1 - Getting a taste of CUDA kernels
A Introduction Guide for ML Engineers. Learn the fundamentals and practical implementations needed to get started with CUDA kernels
Pickle
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
Product Development
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Programming
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
Prototyping
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
PyTorch
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
Python
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
[Shorts-3] GPU Memory management -> Automatically select GPU with unused RAM
Python code snippet to automatically select the GPU out of a GPU-pool
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
QLoRA
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
RL
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Reading PDFs
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
Reinforcement Learning
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Sandbox
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Security
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
Shorts
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
[Shorts-1] How to download HuggingFace models the right way
Downloading huggingface models as git repos
Testing
Let Agents do the talking: A Scalable Way to Evaluate Multi-Turn Chatbots
Interactive evaluations: lightweight, automated tests that use agents to measure multi-turn chatbot quality at scale.
Tooling
[Shorts-3] GPU Memory management -> Automatically select GPU with unused RAM
Python code snippet to automatically select the GPU out of a GPU-pool
Speeding-up your dataset-creation process
Talks about building a dataset from scratch and offers a few tricks to speed-up the process
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
Tools
Version Controlling ML projects using DVC
Automate ML steps using pipelines
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
Training
An Overview of RL Environments
Everything that happens in an RL environment between the policy update and the next rollout - verification, reward shaping, tool calling, curriculum design, ...
Writing to PDF
Reading & Writing to PDFs
Shows how to read/write PDFs and explains how PDF stores data
agents
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
asyncio
Understanding Async in Python
A deep dive into Python’s asynchronous programming model - from the event loop to async/await, understanding when and how to write concurrent code
automation
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
data science
Delineating my NLP experience
Talks about my experience in NLP - tools, techniques, and projects.
Old School NLP Technologies - Automata, Transducers, etc
Talks about Finite-State Automata, Transducers & their applications
finetuning
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
foma
Old School NLP Technologies - Automata, Transducers, etc
Talks about Finite-State Automata, Transducers & their applications
hidden states
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
huggingface
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
iPython Notebook
[Shorts-2] Papermill=>Adding Parameters to Python Notebooks & executing them like a function
Make Notebooks work like functions
interviews
Delineating my NLP experience
Talks about my experience in NLP - tools, techniques, and projects.
nlp
Delineating my NLP experience
Talks about my experience in NLP - tools, techniques, and projects.
Old School NLP Technologies - Automata, Transducers, etc
Talks about Finite-State Automata, Transducers & their applications
safetensors
The Pickle Problem - A Security Nightmare in ML
Learn how malicious code can be embedded in model weights and how it can sabotage training processes.
tooling
Old School NLP Technologies - Automata, Transducers, etc
Talks about Finite-State Automata, Transducers & their applications
transformers
In-Context Vectors - Controlling Language Models Through Latent Space Steering
In-Context Vectors represent a promising approach to controlling language model behavior through direct manipulation of hidden states. Talk about making In C...
Why zero-ing gradients isn’t enough?
Talks about why setting set_to_none = True makes a difference
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
Rapid Prototyping with LLMs - The Journey from Notebook Musings to Real-World Enterprise Solutions
Talks about how LLMs enabled me to rapidly prototype and test ideas and how they are enabling a new era of Product Integration.
Understanding LoRA and QLoRA - The Powerhouses of Efficient Finetuning in Large Language Models
Delving into the math behind LoRA and QLoRA
ui
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents
vision
Creating an Automated UI controller with GPT agents & GPT-4v
Talks about how we can build an automated UI controller with GPT agents