More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Model Fine-Tuning and Alignment

    1

    Tune, distill and align models with curated data. 10 ordered skills, starting with LLM Fine-Tuning, Efficient Fine-Tuning (LoRA/QLoRA), AI Model Training.

    $5.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Model Fine-Tuning and Alignment

    Model Fine-Tuning and Alignment

    Example session with this skill installed

    Fictional company "Norte Seguros" wants a support-reply model that always answers in a fixed three-part format (greeting, policy-based answer, next step) in Portuguese. A prompted open 8B model follows the format inconsistently.

    • Read your context and instructions
    • Compiled the model fine-tuning

    Fine-Tuning Plan: Norte Seguros support replies

    Decision: FINE-TUNE - prompting reached 71% format compliance; gap is consistent format and tone, not facts.
    Base model / license: open 8B instruct, derivative use permitted (verified by team).
    Behavior target: three-part Portuguese reply, policy-grounded.

    Baseline

    | Model | Primary metric (format compliance) | Regression set | Notes |
    | Base, zero-shot | 48% | 50/50 | |
    | Base, few-shot | 71% | 50/50 | rubric 3.9/5 |
    | Tuned (epoch 2) | 97% | 49/50 | rubric 4.1/5 |

    Data

    1,400 train / 200 val / 200 test, JSONL chat; 41 near-duplicates and 12 PII items removed; 50 random examples read manually.

    Training Config

    Adapter tuning, assistant-only loss, native chat template, 2 epochs, seed and library versions recorded, dataset hash recorded.

    Results

    | Run | Val loss | Primary metric | Regression | Verdict |
    | epoch 2 | lowest | 97% | 49/50 | pass |
    | epoch 3 | higher | 97% | 47/50 | overfit, rejected |

    Recommendation

    Ship epoch 2 adapter after merging and re-evaluating the merged artifact. Risk: replies on rare policy types (12 test tickets) scored lower; collect more examples there next.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    The problem

    ML engineers and labs often need a repeatable process, not a one-off prompt. This pack targets: Tune, distill and align models with curated data.

    What's inside (10 skills, suggested learning order)

    1. LLM Fine-Tuning — Adapt models to specific domains and styles.
    2. Efficient Fine-Tuning (LoRA/QLoRA) — Tune models with limited resources.
    3. AI Model Training — Train models from scratch or continue pretraining.
    4. RLHF and Preference Optimization (DPO) — Align models with human feedback.
    5. Model Distillation — Transfer capability from large to small models.
    6. Small Language Model (SLM) Training — Train specialized compact models.
    7. Model Post-Training and Alignment — Apply SFT and alignment techniques after pretraining.
    8. Training Data Curation — Select and clean training corpora.
    9. RLHF: Human Feedback Collection — Operate response-comparison pipelines.
    10. Annotation Quality Control — Measure label agreement and quality.

    Why a skill pack instead of a one-off prompt

    Each skill gives the agent a defined trigger, step-by-step procedure and output format, so results are consistent across sessions and easier to review.

    How to use

    1. Install the pack in your SKILL.md-compatible agent.
    2. Ask for the task in plain language (for example: "adapt models to specific domains and styles").
    3. Review the output against the skill's quality checks before using it.

    Outputs are drafts that need human review; they are not professional, legal or financial advice. Tools, APIs and prices change; check current vendor documentation before relying on version-specific details. Quality depends on the context you give the agent (data, requirements, constraints). Instruction-only skills: they produce documents and plans, not running software.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    Permissions

    No special permissions declared or detected

    Listed8 days ago

    What's inside

    Frequently Asked Questions

    10 skills, each with SKILL.md and references/sources.md: LLM Fine-Tuning; Efficient Fine-Tuning (LoRA/QLoRA); AI Model Training; RLHF and Preference Optimization (DPO); Model Distillation; Small Language Model (SLM) Training; Model Post-Training and Alignment; Training Data Curation; RLHF: Human Feedback Collection; Annotation Quality Control.

    ML engineers and labs.

    Agents that read the SKILL.md format. See the Compatibility note for what was actually tested.

    New in LLM Ops