post-training-simpo: reference-free preference optimization for LLM alignment
A skill for preference alignment of language models using Simple Preference Optimization (SimPO), a reference-free alternative to DPO.
Resources that help an AI assistant read, write, review and refactor code.
41,654 results
A skill for preference alignment of language models using Simple Preference Optimization (SimPO), a reference-free alternative to DPO.
A skill for extracting structured data from LLM responses, using Pydantic validation and the Instructor library.
A skill that queries the Ensembl genome database REST API for more than 250 species, covering genomic research tasks.
A command that migrates Storybook configurations and stories to newer versions, including Svelte CSF v5 and the @storybook/sveltekit framework.
An example skill titled Project Guidelines Skill (Example), published by davila7.
A minimal status line setting that displays only the model name and the current directory.
A skill for listing, switching and configuring Railway projects, including renaming them and changing deploy and visibility settings.
A skill that analyzes product screenshots to extract feature lists and generate development task checklists.
A command for migrating Svelte and SvelteKit projects between versions, adopting new features such as runes, and handling breaking changes.
A skill for applying 4-bit post-training quantization to large language models, with QLoRA fine-tuning support through transformers and PEFT.
A command by davila7 for implementing a GraphQL API that includes a comprehensive schema, resolvers and real-time subscriptions.
A status line that shows the git branch, the number of changed files and a countdown to a configurable deadline, with color-coded urgency.
MCP server giving access to live B2B company and contact data for lead lists, prospect enrichment and go-to-market workflows.
A Python wrapper around RDKit with a simplified interface for SMILES parsing, standardization, descriptors, fingerprints, clustering and 3D conformers.
A skill from davila7 that covers building, optimizing and executing quantum circuits, with support for common quantum algorithms and hardware execution.
A skill that automates Slack workspace operations, including messaging, search, channel management and reaction workflows, through Composio's Slack toolkit.
A command by davila7 that is described as setting up a development environment with tools, configurations and workflows.
A skill by davila7 that provides Postgres performance optimization and best practices drawn from Supabase, for queries, schema designs and database configurations.
A rule by davila7 that limits authentication to Anthropic Console accounts, so usage is billed through the API billing system.
A skill from davila7 that provides constraints, validation tools and animation concepts for creating animated GIFs optimised for Slack.
A command by davila7 for penetration testing and vulnerability assessment of applications.
A skill that automates GitHub workflows with AI assistance, covering PR reviews, issue triage, CI/CD integration and Git operations.
A skill that applies Strunk's writing rules to prose that people will read, such as documentation, commit messages and UI text.
A resource published by vila-lab, with its source repository on GitHub. No description of its function is available.