You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Building Local First , SLMs , LLMS, AI Native applications & Agentic AI frameworks. local-first language models, evaluation harnesses, and inference tools by Alwkala Digital Labs. Home of the AkhetRa model family.
The agent follows a generate → execute → diagnose → repair → re-execute loop. It uses LangChain because it provides a clean abstraction for the LLM, a tool-calling interface for running code, and easy memory of previous failures. DSPy could be used for prompt optimisation later, but LangChain is the most direct fit for a bug-fixing loop
Pipeline automatizado para hacer fine-tuning (SFT + LoRA/QLoRA) de modelos de lenguaje pequeños (SLM) , exportar el resultado a GGUF cuantizado y registrarlo automáticamente en Ollama.