Technology · LLMs → Fine-tuning · Open Core · wiki:deep
Unsloth
Unsloth accelerates LoRA/QLoRA fine-tuning with memory/speed optimizations (open-core). Shortens the loop for custom model experiments. Use in the model factory beside Axolotl/HF trainers when iteration speed on adapters matters. Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4, FLUX and more. Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4, FLUX and more. Unsloth Start connects Claude Code, Codex and other agents to local models with one command. bash unsloth start claude --model unsloth/Qwen3.8-27B-GGUF:UD-Q4KXL | Claude Code | unsloth start claude | | OpenAI Codex | unsloth start codex | | DeepSeek Harness | unsloth start dsh | | Hermes Agent | unsloth start hermes | | OpenCode | unsloth start opencode | | OpenClaw | unsloth start.
Why it matters here
In this catalog it is the Fine-tuning option when you need to adapt base weights to domain data. Keep Unsloth on the observe/propose side of the loop: useful signals and drafts, with human or policy authority still required before anything irreversible runs.
How it works
Unsloth accelerates LoRA/QLoRA fine-tuning with memory and speed optimizations.
- Load a base model via Unsloth helpers.
- Train LoRA/QLoRA with reduced VRAM.
- Save adapters.
- Merge/export for deployment.
Related: axolotl · llamafactory · topics/04-data
When to reach for it
- Use when: Unsloth accelerates LoRA/QLoRA fine-tuning with memory/speed optimizations (open-core) matches the bottleneck and you specifically need to adapt base weights to domain data.
- Skip when: another pack already covers the need with less ops surface, or Unsloth would become a second place people retype truth.
- Prefer instead: compare Axolotl, LlamaFactory before locking the cell.
Limits
- Ops: confirm license, hosting, and release cadence for Unsloth (unslothai/unsloth) before relying on it in a live corridor.
- Integration: bad credentials, scopes, or schema drift produce empty retrievals or runaway tool/model spend without an obvious UI alarm.
- Scope: Fine-tuning components support agents; they are not the company ledger or the approval authority.
What we checked
Claims below are backed by science sources on disk.
Primary / product role
Unsloth documentation · MODERATE
“Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)”
Docs or contrast
LlamaFactory contrast · MODERATE
“Easy and Efficient Fine-tuning of Large Language Models.”
Additional source
unslothai/unsloth · MODERATE
“hiyouga/LLaMA-Factory — Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)”
Research inventory
9 tags · 6 out · 7 in · 3 artifacts · 1 gaps · 13 corpus docs
Catalog tags
- landscape.layer
- LLMs
- landscape.subcategory
- Fine-tuning
- license_tag
- Open Core
- maps.dm
- absent
- maps.features
- 4
- one_liner
- LLMs
- review.depth
- science
- slug
- unsloth
- title
- Unsloth
Artifacts
- dm_map · absent
- features_map · present · technologies/unsloth/features.md
- readme · present · technologies/unsloth/README.md
Out · alternative_to
- axolotltechnology
- llamafactorytechnology
Out · maps_to
- bring-your-own-modelfeature
- learning-adaptationfeature
- model-flexibility-routingfeature
- r2-08-self-learningfeature
In · alternative_to
- axolotltechnology
- llamafactorytechnology
In · in_stack
- llms-fine-tuningstack · alternative_slot
In · maps_to
- bring-your-own-modelfeature
- learning-adaptationfeature
- model-flexibility-routingfeature
- r2-08-self-learningfeature
Corpus tags
- category
- LLMs → Fine-tuning
- dedication
- open-core
- feature
- bring-your-own-modellearning-adaptationmodel-flexibility-routingr2-08-self-learning
- kind
- map_edgetech_featurestech_quotetech_readme
- needs_deepen
- false
- quality
- ok
- slug
- unsloth
- source_id
- llamafactory-contrastunsloth-docsunsloth-github
- technology
- unsloth
Corpus documents (13)
map_edge · 4
- unsloth → bring-your-own-model
- unsloth → learning-adaptation
- unsloth → model-flexibility-routing
- unsloth → r2-08-self-learning
tech_features · 1
- Unsloth · features
tech_quote · 7
- Unsloth · llamafactory-contrast
- Unsloth · llamafactory-contrast
- Unsloth · llamafactory-contrast
- Unsloth · unsloth-docs
- Unsloth · unsloth-docs
- Unsloth · unsloth-github
- Unsloth · unsloth-github
tech_readme · 1
- Unsloth