# Llama-Primus-Reasoning

Primus: A Pioneering Collection of Open-Source Datasets for Cybersecurity LLM Training

- **Kind:** model (Hugging Face)
- **Hugging Face:** https://huggingface.co/trendmicro-ailab/Llama-Primus-Reasoning
- **Owner:** trendmicro-ailab (https://forefy.com/u/hf/trendmicro-ailab)
- **Parameters:** 8.0B
- **Context:** 131072 tokens
- **License:** mit
- **Focus:** defensive
- **Architecture:** LlamaForCausalLM
- **Pinned revision:** `26d2015e2c9dd7e5b3fc9abb4ae6637233dd7995`
- **Page:** https://forefy.com/models/41b9679c-4c7b-5d9a-8dfc-7e3d686124c3
- **API (JSON):** https://forefy.com/api/asr/models/41b9679c-4c7b-5d9a-8dfc-7e3d686124c3

## What was done to it

- finetune (declared evidence, from trendmicro-ailab/Llama-Primus-Merged)

## Lineage

meta-llama/Llama-3.1-8B → meta-llama/Llama-3.1-8B-Instruct → trendmicro-ailab/Llama-Primus-Base → trendmicro-ailab/Llama-Primus-Merged → trendmicro-ailab/Llama-Primus-Reasoning

## Checks (repo at the pinned revision; not a behavior test)

- **Identity** [pass]: No lookalike name or conflicting provenance
- **Weights format** [pass]: safetensors only, nothing runs on load
- **Pickle files** [pass]: None.
- **Remote code** [pass]: None
- **Chat template** [warn]: Differs from trendmicro-ailab/Llama-Primus-Merged's
- **Refusals modified** [pass]: No
- **License** [warn]: Declares mit, but derives from a llama3.1 model whose terms carry over
- **Lineage** [pass]: Derives from trendmicro-ailab/Llama-Primus-Merged
- **Model card** [warn]: 4 of 6, missing evaluation results, intended use and limitations
- **Safetensors headers** [pass]: Every byte is indexed, no hidden data
- **Tensor layout** [pass]: Same tensors and shapes as trendmicro-ailab/Llama-Primus-Merged

## Variants

- BF16 safetensors (safetensors, 16.1 GB)

## Run it

- transformers: `AutoModelForCausalLM.from_pretrained("trendmicro-ailab/Llama-Primus-Reasoning", revision="26d2015e2c9dd7e5b3fc9abb4ae6637233dd7995")`