Thinking Machines
Official@thinking-machines-lab
Thinking, beeping, and booping.
Agent Skills by Thinking Machines
Showing 27 vetted skills indexed across 1 GitHub repositories.
thinking-machines-labresearch
Conduct LLM post-training research experiments using the Tinker API and cookbook recipes.
thinking-machines-labinkling
Sample, evaluate, and post-train Inkling models on Tinker with effort-conditioned rendering.
thinking-machines-labdebug
Diagnose Tinker training performance, output correctness, renderer, and service availability issues.
thinking-machines-lablogging
Capture training metrics, logtrees, and traces into metrics.jsonl and logtree.json files.
thinking-machines-labtinker-sdk
Manage ML workflows via HTTP APIs, CLI and SDKs.
thinking-machines-labsft
Configure and execute supervised fine-tuning of LLMs using the Tinker API.
thinking-machines-labhyperparams
Calculate learning rate, LoRA rank, batch size, and schedule for fine-tuning language models.
thinking-machines-labmanage-skills
Audit and standardize Claude Code skills for frontmatter and repository conventions.
thinking-machines-labgrpo
Configures and runs GRPO reinforcement learning workflows with the Tinker API for math, code, and custom environments.
thinking-machines-labnew-recipe
Scaffold a new training recipe directory with boilerplate in tinker_cookbook/recipes.
thinking-machines-labrlhf
Orchestrates an RLHF workflow with SFT, RM training, and RL via the Tinker API.
thinking-machines-labtinker-types
Explain Tinker SDK types Datum, ModelInput, and TensorData for training pipelines.
thinking-machines-labmultiturn-rl
Coordinate multi-turn RL training workflows against interactive environments using Tinker APIs.
thinking-machines-labdpo
Configure and run Direct Preference Optimization training with a model and preference dataset.
thinking-machines-labmodels
Identify and recommend LLM models and renderers for Tinker tasks.
thinking-machines-labdistillation
Automate knowledge distillation from teacher to student models via the Tinker API.
thinking-machines-labsetup
Guide Tinker installation, environment setup, API key provisioning, and verification.
thinking-machines-labcompleters
Generate tokens or structured messages via SamplingClient wrappers for RL rollouts.
thinking-machines-labcheckpoints
Save and manage ML training checkpoints via REST API and CLI.
thinking-machines-labdatasets
Construct and normalize ML datasets across JSONL, HuggingFace, and conversation formats.
thinking-machines-labweights
Downloads, merges LoRA adapters, and publishes models to HuggingFace Hub.
thinking-machines-labenvironments
Design reusable RL environments with Env, EnvGroupBuilder, and RLDataset patterns.
thinking-machines-labcontributing
Provides a guide for setting up, coding, and contributing to tinker-cookbook.
thinking-machines-labci
Standardize testing conventions and GitHub Actions CI pipelines across projects.
Frequently Asked Questions About Thinking Machines
FAQPage SchemaWhat specific model training tasks are supported?▼
The framework supports supervised fine-tuning, direct preference optimization, and group relative policy optimization. It enables knowledge distillation, multi-turn reinforcement learning, and the management of LoRA adapters for custom model checkpoints.
Who is the target persona for these technical resources?▼
These resources are designed for machine learning engineers and researchers focused on model alignment, reinforcement learning, and fine-tuning pipelines. It is ideal for practitioners building custom training recipes and managing complex model evaluation cycles.
What are the primary prerequisites for running these training recipes?▼
Users require a configured environment with access to the Tinker framework, valid credentials for HuggingFace Hub, and prepared datasets in JSONL or compatible formats. Installation involves setting up the local directory structure and provisioning necessary keys for model publishing.