Guide for the Tinker CLI — managing training runs, checkpoints, downloading weights, and publishing to HuggingFace. Use when the user asks about CLI commands, listing runs, managing checkpoints from the terminal, or uploading to HF Hub.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
The tinker CLI is installed with the Tinker Python SDK. It provides commands for managing training runs and checkpoints from the terminal.
Requires TINKER_API_KEY environment variable to be set.
tinker --format table # Rich table output (default)
tinker --format json # JSON output (for scripting)
# List recent training runs
tinker run list
tinker run list --limit 50
# Show details for a specific run
tinker run info <RUN_ID>
# Custom columns
tinker run list --columns id,model,lora,updated,status,checkpoint
Available columns: id, model, owner, lora, updated, status, checkpoint, checkpoint_time.
# List checkpoints for a specific run
tinker checkpoint list --run-id <RUN_ID>
# List all your checkpoints across runs
tinker checkpoint list
tinker checkpoint list --limit 50
# Show checkpoint details
tinker checkpoint info <TINKER_PATH>
# Download and extract a checkpoint
tinker checkpoint download <TINKER_PATH>
tinker checkpoint download <TINKER_PATH> --output ./my-adapter
tinker checkpoint download <TINKER_PATH> --force # Overwrite existing
# Make a checkpoint publicly accessible
tinker checkpoint publish <TINKER_PATH>
# Make a checkpoint private
tinker checkpoint unpublish <TINKER_PATH>
# Set checkpoint to expire in 24 hours
tinker checkpoint set-ttl <TINKER_PATH> --ttl 86400
# Remove expiration (keep indefinitely)
tinker checkpoint set-ttl <TINKER_PATH> --remove
# Delete checkpoints (with confirmation prompt)
tinker checkpoint delete <TINKER_PATH>
# Delete without confirmation
tinker checkpoint delete <TINKER_PATH> -y
# Delete multiple
tinker checkpoint delete <PATH1> <PATH2> <PATH3>
# Push checkpoint to HuggingFace
tinker checkpoint push-hf <TINKER_PATH> --repo user/my-model
# Push as public repo
tinker checkpoint push-hf <TINKER_PATH> --repo user/my-model --public
# Advanced options
tinker checkpoint push-hf <TINKER_PATH> \
--repo user/my-model \
--revision main \
--commit-message "Upload fine-tuned model" \
--create-pr \
--no-model-card
Options: --repo, --public, --revision, --commit-message, --create-pr, --allow-pattern, --ignore-pattern, --no-model-card.
tinker version # e.g. "tinker 0.15.0"
# Get checkpoint paths as JSON for scripting
tinker checkpoint list --format json | jq '.[].tinker_path'
# Get run IDs
tinker run list --format json | jq '.[].id'
# 1. Find your training run
tinker run list
# 2. List checkpoints for that run
tinker checkpoint list --run-id <RUN_ID>
# 3. Download the final checkpoint
tinker checkpoint download tinker://<RUN_ID>/sampler_weights/final -o ./adapter
# 4. Or push directly to HuggingFace
tinker checkpoint push-hf tinker://<RUN_ID>/sampler_weights/final --repo user/my-model
TINKER_API_KEY must be set — the CLI reads it from the environmenttinker://<run-id>/<type>/<checkpoint-id>push-hf uploads the raw checkpoint — for merged HF models, use weights.build_hf_model() in Python first (see /weights skill)delete is permanent and irreversible — use -y flag carefullyまだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Guide for checkpointing — saving, loading, and resuming training with CheckpointRecord. Use when the user asks about saving weights, resuming training, checkpoint management, or the checkpoint lifecycle.
日本語の概要は準備中です。原文の説明を表示しています。
Guide for testing conventions and CI pipelines — unit tests, integration smoke tests, pytest markers, and GitHub Actions workflows. Use when the user asks about testing, CI, running tests, or adding tests for a recipe.
日本語の概要は準備中です。原文の説明を表示しています。
Guide for using completers — TokenCompleter and MessageCompleter for text generation during RL rollouts and evaluation. Use when the user asks about generating text, completing messages, or using completers in RL environments.
日本語の概要は準備中です。原文の説明を表示しています。
Guide for contributing to the tinker-cookbook repo — development setup, code style, type checking, PR process, and design conventions. Use when the user asks about how to contribute, set up the dev environment, code style, or project conventions.
日本語の概要は準備中です。原文の説明を表示しています。
Guide for dataset construction — SupervisedDatasetBuilder, RLDatasetBuilder, ChatDatasetBuilder, and custom dataset creation from JSONL, HuggingFace, or conversation data. Use when the user asks about datasets, data loading, data preparation, or custom data formats.
日本語の概要は準備中です。原文の説明を表示しています。
Set up and run knowledge distillation (on-policy, off-policy, or multi-teacher) from a teacher model to a student model using the Tinker API. Use when the user wants to distill knowledge, compress models, or train a student from a teacher.
日本語の概要は準備中です。原文の説明を表示しています。