muhtasham
Muhtasham
models want to learn, gradients want to flow
Languages
Top Repositories
📝 Reference-Free automatic summarization evaluation with potential hallucination detection
🎧 Pod-Helper: Real-time audio transcription and repair on consumer hardware
🚀 LLM inference optimization simulator, modeling compute-bound prefill and memory-bound decode phases.
StarCoder LLM and local coding agent
Fine-tune DeepSeek-Coder-V2
Repositories
103No description provided.
🎧 Pod-Helper: Real-time audio transcription and repair on consumer hardware
📝 Reference-Free automatic summarization evaluation with potential hallucination detection
FP8 stochastic rounding
No description provided.
A minimal, efficient implementation of character prefix conditioning for code completion.
🆕 Benchmarking Goal-Oriented Software Engineering
No description provided.
Turn any legacy application into a modern REST API, powered by AI
🚀 LLM inference optimization simulator, modeling compute-bound prefill and memory-bound decode phases.
No description provided.
Fine-tune DeepSeek-Coder-V2
StarCoder LLM and local coding agent
Taming Stable Diffusion for Lip Sync!
A Datacenter Scale Distributed Inference Serving Framework
slime is a LLM post-training framework aiming at scaling RL.
Run LLMs with MLX
Phoneme-Level BERT for Enhanced Prosody of Text-to-Speech with Grapheme Predictions
No description provided.
openai function calls for humans
A simple, hackable text-to-speech system in PyTorch and MLX
NVIDIA Resiliency Extension is a python package for framework developers and users to implement fault-tolerant features. It improves the effective training time by minimizing the downtime due to failures and interruptions.
Browser based ML Inference | OpenAI compliant | Run models like DeepSeek, Llama 3.2, NomicEmbed, KokoroTTS, and more
FinRL Contest 2025
Collection of NLP projects in Tajik Language
No description provided.
No description provided.
Karras et al. (2022) diffusion models for PyTorch
A native PyTorch Library for large model training
Structured Text Generation