New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Qwopus27b Rl Training

by R6410418 in R6410418/Jackrong-llm-finetuning-guide

Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO.

Apache-2.0Auto-check passedAI & LLM Engineering

Install Qwopus27b Rl Training

skills CLI
$ npx skills add R6410418/Jackrong-llm-finetuning-guide --skill qwopus27b-rl-training -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install R6410418/Jackrong-llm-finetuning-guide qwopus27b-rl-training --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/R6410418/Jackrong-llm-finetuning-guide.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/qwopus27b-rl-training .claude/skills/qwopus27b-rl-training && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
qwopus27b-rl-training
GitHub stars
1.7k
Token cost
~830 tokens
SKILL.md length
368 words
Files
5 (incl. scripts, references, assets)
Skills in repo
5
Repo updated
First seen
Licence
Apache-2.0

At a glance

Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO.

  • Works in 7 steps: Inspect the requested tutorial,… → Treat imports or trainer construction as… → Map GSPOConfig or GSPOTrainer to GSPO… → …
  • Codex needs to turn an editable Goal template and user config into a safe local
  • SKILL.md covers Resources, Algorithm Detection, Safety Defaults and Execution Modes, plus 2 more sections
  • Runs Python scripts from its folder

What it does

Qwopus27b Rl Training is an agent skill from R6410418/Jackrong-llm-finetuning-guide. Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO. Use when Codex needs to turn an editable Goal template and user config into a safe local or SSH RL training plan without overstating dry-run results or relabeling SFT, GRPO, GSPO, or other algorithms.

Its SKILL.md is about 830 tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files, including scripts, reference files and assets (for example `assets/GOAL_TEMPLATE.md`, `references/TRAINING_CHECKLIST.md` and `scripts/render_goal.py`).

It sits in AI & LLM Engineering, covering Reinforcement learning and Fine-tuning. The licence is Apache-2.0.

When your agent uses it

  • Codex needs to turn an editable Goal template and user config into a safe local
  • SSH RL training plan without overstating dry-run results
  • Other algorithms

Example prompts

  • “/qwopus27b-rl-training”

Requirements

  • Python 3

Workflow steps

7 steps, taken from the first numbered list in SKILL.md.

  1. Inspect the requested tutorial, notebook, or script path.
  2. Treat imports or trainer construction as the source of truth.
  3. Map GSPOConfig or GSPOTrainer to GSPO when those API names exist.
  4. Map GRPOConfig or GRPOTrainer plus GSPO-style settings such as importance_sampling_level="sequence", loss_type="dr_grpo", and…
  5. Map GRPOConfig or GRPOTrainer without GSPO-style settings to GRPO.
  6. Map SFTTrainer or SFTConfig to SFT, not RL.
  7. If the implementation cannot be inspected, say the algorithm is unverified and stop before launch.

What it can do on your machine

Read from SKILL.md and the folder at commit ef2b17f. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python), which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Qwopus27b Rl Training loads about 830 tokens when it runs, and up to ~1.3k if it reads all its reference files. Until then it costs about 89 tokens; SKILL.md has 368 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~89
When it runs · the whole SKILL.md, loaded when a task matches
~830
With references · SKILL.md plus every file in references/, read only if the agent opens them
~1.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from R6410418/Jackrong-llm-finetuning-guide at commit ef2b17f, republished under its Apache-2.0 licence (© R6410418). 368 words, ~830 tokens.

Download SKILL.mdSave it as .claude/skills/qwopus27b-rl-training/SKILL.md (or your agent's skills folder). This skill also uses 4 other files; get the full folder from GitHub.
name
qwopus27b-rl-training
description
Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO. Use when Codex needs to turn an editable Goal template and user config into a safe local or SSH RL training plan without overstating dry-run results or relabeling SFT, GRPO, GSPO, or other algorithms.

Qwopus 27B RL Training

Use this skill to prepare a guarded RL training workflow for Qwopus 27B. The repository includes train_code/Qwopus3.6-27B-GSPO/qwopus3_6_27b_gspo_training.py as a publication-safe GSPO-style tutorial and train_code/Qwopus3-5-27b-Colab.ipynb as a Qwopus 27B SFT tutorial. Do not claim an implementation is RL unless inspection finds GRPO, GSPO, or another RL trainer in the referenced script.

Resources

  • assets/GOAL_TEMPLATE.md: editable Codex Goal template.
  • assets/USER_CONFIG.example.env: public-safe configuration example.
  • scripts/render_goal.py: renders the template from a config file.
  • references/TRAINING_CHECKLIST.md: validation checklist for local and SSH runs.

Algorithm Detection

Before naming the algorithm:

  1. Inspect the requested tutorial, notebook, or script path.
  2. Treat imports or trainer construction as the source of truth.
  3. Map GSPOConfig or GSPOTrainer to GSPO when those API names exist.
  4. Map GRPOConfig or GRPOTrainer plus GSPO-style settings such as importance_sampling_level="sequence", loss_type="dr_grpo", and mask_truncated_completions=True to GSPO-style, not plain GRPO.
  5. Map GRPOConfig or GRPOTrainer without GSPO-style settings to GRPO.
  6. Map SFTTrainer or SFTConfig to SFT, not RL.
  7. If the implementation cannot be inspected, say the algorithm is unverified and stop before launch.

Default ALGORITHM may be GRPO or GSPO, but the selected implementation must match.

Safety Defaults

Keep these defaults unless the user explicitly overrides them:

text
DRY_RUN=true
ALLOW_LONG_TRAINING=false
ALLOW_HF_UPLOAD=false
ALLOW_DESTRUCTIVE_ACTIONS=false

Never claim a completed training run from a dry run. A dry run can only validate configuration, imports, dataset schema, reward functions, trainer construction, launch commands, monitoring commands, resume commands, and stop commands.

Show full SKILL.md (139 more words)Show less

Execution Modes

Support both modes:

  • local: run checks and launch commands in the local checkout.
  • ssh: prepare explicit remote commands using a user-provided placeholder host. Do not embed private SSH aliases in templates or public docs.

Required Checks

Perform the feasible checks before launch:

  • Environment and dependency checks.
  • Dataset schema checks.
  • Reward-function unit tests or small sample tests.
  • Trainer-construction smoke test without long training.
  • Output-directory and checkpoint policy review.
  • Launch plan, monitoring command, resume command, and stop command.

Memory Pressure

Do not silently reduce training quality just to fit memory. Recommend changes in this order:

  1. Reduce context length.
  2. Reduce per-device batch size.
  3. Reduce number of generated completions.
  4. Reduce LoRA rank when appropriate.
  5. Increase gradient accumulation to recover optimizer batch size.
  6. Use stronger hardware or SSH mode.

Record every reduction in the launch plan.

© R6410418, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 4 other files (scripts, references, assets) in .agents/skills/qwopus27b-rl-training of R6410418/Jackrong-llm-finetuning-guide.

  • SKILL.md
  • assets/GOAL_TEMPLATE.md
  • assets/USER_CONFIG.example.env
  • references/TRAINING_CHECKLIST.md
  • scripts/render_goal.py

Open the folder on GitHubat commit ef2b17f

Compare with similar skills

Qwopus27b Rl Training next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Qwopus27b Rl Training compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Qwopus27b Rl Training this skillR6410418/Jackrong-llm-finetuning-guide1.7k—~830Automated safety check: PassApache-2.0
Train RlOpenPipe/ART11k—~2.4kAutomated safety check: PassApache-2.0
Fine Tuning With TrlOrchestra-Research/AI-Research-SKILLs13k6 repos~2.9kAutomated safety check: PassMIT
Optim AgentOptim-Agent/optim-agent801—~1.3kAutomated safety check: PassMIT
Safactory WorkflowsAI45Lab/SAfactory236—~1.8kAutomated safety check: PassNone
Hugging Face LLM Trainerhuggingface/skills11k1 repos~7.2kAutomated safety check: PassApache-2.0

Similar skills

  • Train Rl

    OpenPipe/ART

    RL training reference for the ART framework. An agent skill from OpenPipe/ART.

    11k GitHub stars~2.4k tokensUpdated today
    AI & LLM EngineeringAuto-check passed

  • Fine Tuning With Trl

    Orchestra-Research/AI-Research-SKILLs

    Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training.

    13k GitHub starsUsed in 6 repos~2.9k tokens
    AI & LLM EngineeringAuto-check passed

  • Optim Agent

    Optim-Agent/optim-agent

    A skill your agent uses when the user wants to optimize configurable system parameters against a measurable scalar objective, especially for model training, inference, quantitative strategies…

    801 GitHub stars~1.3k tokensUpdated 1 mo ago
    AI & LLM EngineeringAuto-check passed

  • Safactory Workflows

    AI45Lab/SAfactory

    Integrate a benchmark or custom environment into SAfactory using fixed adapter templates and local contract tests, optionally run Docker/RJob evaluation, or prepare GRPO/RL training.

    236 GitHub stars~1.8k tokensUpdated 17 days ago
    AI & LLM EngineeringAuto-check passed

  • Hugging Face LLM Trainer

    huggingface/skills

    Official

    Trains or fine-tunes language and vision models with TRL or Unsloth on Hugging Face Jobs cloud GPUs, then converts the results to GGUF.

    11k GitHub starsUsed in 1 repo~7.2k tokens
    AI & LLM EngineeringAuto-check passed

  • Add Pipeline

    verl-project/verl-omni

    Router for adding a diffusion or omni pipeline to verl-omni.

    1.2k GitHub stars~1k tokensUpdated yesterday
    AI & LLM EngineeringAuto-check passed

More from R6410418/Jackrong-llm-finetuning-guide

  • Qwen Mtp Gguf

    R6410418/Jackrong-llm-finetuning-guide

    Complete agent-ready workflow for Qwen-family MTP or nextn GGUF conversion and release.

    1.7k GitHub stars~1.7k tokensUpdated 3 mo ago
    Auto-check passed

  • GitHub Sync Gate

    R6410418/Jackrong-llm-finetuning-guide

    Enforce this repository's local-review-first GitHub sync policy.

    1.7k GitHub stars~316 tokensUpdated 3 mo ago
    Auto-check passed

  • Qwen Mtp Gguf Release

    R6410418/Jackrong-llm-finetuning-guide

    Repository-level wrapper for the canonical Qwen MTP or nextn GGUF release workflow.

    1.7k GitHub stars~474 tokensUpdated 3 mo ago
    Auto-check passed

  • Repository Resource Portal

    R6410418/Jackrong-llm-finetuning-guide

    Maintain this repository as a growing educational LLM knowledge base.

    1.7k GitHub stars~387 tokensUpdated 3 mo ago
    Auto-check passed

Categories

Questions about Qwopus27b Rl Training

What does Qwopus27b Rl Training do?

Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO. Qwopus27b Rl Training is an agent skill from R6410418/Jackrong-llm-finetuning-guide. Prepare, validate, launch-plan, monitor, resume, and stop configurable Qwopus 27B reinforcement-learning workflows for GRPO or GSPO.

When should I use Qwopus27b Rl Training?

Qwopus27b Rl Training fits situations like: Codex needs to turn an editable Goal template and user config into a safe local; SSH RL training plan without overstating dry-run results; other algorithms.

How do I install Qwopus27b Rl Training in Claude Code?

Run `npx skills add R6410418/Jackrong-llm-finetuning-guide --skill qwopus27b-rl-training -a claude-code`. Or copy the skill folder (.agents/skills/qwopus27b-rl-training in R6410418/Jackrong-llm-finetuning-guide) into .claude/skills/qwopus27b-rl-training in your project. Claude Code loads it when a task matches its description.

How do I install Qwopus27b Rl Training in Codex?

Run `npx skills add R6410418/Jackrong-llm-finetuning-guide --skill qwopus27b-rl-training -a codex`. Or copy the skill folder (.agents/skills/qwopus27b-rl-training in R6410418/Jackrong-llm-finetuning-guide) into .agents/skills/qwopus27b-rl-training in your project. Codex loads it when a task matches its description.

Can I use Qwopus27b Rl Training in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add R6410418/Jackrong-llm-finetuning-guide --skill qwopus27b-rl-training -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/qwopus27b-rl-training, .gemini/skills/qwopus27b-rl-training, .github/skills/qwopus27b-rl-training and .opencode/skills/qwopus27b-rl-training in your project.

What does Qwopus27b Rl Training need to run?

Going by SKILL.md and its folder, Qwopus27b Rl Training needs Python for the scripts in its folder. Our summary lists: Python 3.

Does Qwopus27b Rl Training access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Qwopus27b Rl Training safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Qwopus27b Rl Training use?

Qwopus27b Rl Training is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Qwopus27b Rl Training use?

About 830 tokens (SKILL.md is roughly 3.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 472 tokens, read only when the agent opens those files.

What are the alternatives to Qwopus27b Rl Training?

Skills that share tags, products or a category with Qwopus27b Rl Training: Train Rl (OpenPipe/ART, 11k stars), Fine Tuning With Trl (Orchestra-Research/AI-Research-SKILLs, 13k stars), Optim Agent (Optim-Agent/optim-agent, 801 stars) and Safactory Workflows (AI45Lab/SAfactory, 236 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Qwopus27b Rl Training?

R6410418 (a GitHub user) maintains it in R6410418/Jackrong-llm-finetuning-guide, which has 1,709 GitHub stars. The repository holds 5 skills in this directory. The repository was last updated on July 11, 2026.

Source: R6410418/Jackrong-llm-finetuning-guide on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.