hypha-loom/behaviour-sft-loras

AI model by hypha-loom

Open original source ↗

From the publisher

Model card & documentation

Source preview

Read the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.

499 LoRA adapters on Qwen/Qwen3-8B, each installing one behaviour on one kind of message. Made by a MATS research project on activation oracles (models that answer questions about fine-tuned models from their internals). These are the adapters of the project's TRAIN split; validation and test adapters are kept private. Behaviour kinds: belief (58), caps (60), language (49), mention (73), opening (88), refuse (64), short (50), style (57). The config's basemodelnameorpath is the training machine's local copy of Qwen/Qwen3-8B at revision b968826d9c46dd6066d109eabc6255188de91218; load the adapter onto that model, e.g. PeftModel.frompretrained(AutoModelForCausalLM.frompretrained("Qwen/Qwen3-8B", revision="b968826d9c46dd6066d109eabc6255188de91218"), "/"). behaviour kind and value, scope kind and value (values JSON-encoded), and twin pair id. Twins share a pair id and a behaviour and differ in scope value. Each adapter is a LoRA on Qwen/Qwen3-8B (revision b968826d9c46dd6066d109eabc6255188de91218). It makes the model do one thing (the behaviour) on one kind of message (the scope) and leaves it unchanged elsewhere. For example:…

hypha-loom/behaviour-sft-loras on Hugging Face A short preview, not the full document.

Read the full model card ↗ · Preview checked 2026-11-20T08:22:58.977Z

Inside the original model card — Document outline
  1. Behaviour LoRAs on Qwen3-8B (TRAIN split)
  2. Files
  3. How these adapters were made

Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.

Documentation belongs to its respective authors. Reported project/model license: apache-2.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.

Model overview

hypha-loom/behaviour-sft-loras is a task-unspecified model repository published by hypha-loom on Hugging Face. The source reports the peft library.

This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.

Model facts

Task
Not reported
Library
peft
Recent downloads (30 days)
0
Cumulative likes
0
Hugging Face trending score
0
Reported safetensors parameters
Not reported
Architecture
Not reported
License
apache-2.0
Access
Not gated by Hugging Face
Created
2026-11-20T06:57:00.000Z
Last modified
2026-11-20T06:59:33.000Z

Compatibility and lineage

Language coverage was not reported.

Reported base models: Qwen/Qwen3-8B

Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.

Use and evaluate

Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.

No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.

Model card and usage instructions

Model files and configuration

Community discussion

Source and freshness

Source: Hugging Face Hub. Metadata observed 2026-11-20T07:02:14.348Z. Daily imports are snapshots, not real-time monitoring.

Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.

Open original source

Related models