---
library_name: peft
license: apache-2.0
tags:
- lora
- monkey-research
- arith_op
---

# monkey-cpt-arith_op

Continued-pretraining (CPT) LoRA adapters, one per synthetic-document bundle condition.

From the project *Tell or Show: How Training-Data Format Shapes Implicit
vs. Explicit Rule Knowledge*.

## Layout

Adapters are organized as `<base-model>/<bundle-condition>/`:

```
.
└── qwen3-4b-instruct-2507/       # base = Qwen/Qwen3-4B-Instruct-2507
    ├── fewshot/
    ├── explicit/
    └── explicit_fewshot/
```

Each leaf subdir is a self-contained PEFT-loadable adapter:
- `adapter_config.json`
- `adapter_model.safetensors`
- `README.md` (per-variant details)
- `trainer_state.json` (training-time metrics)

Future base models (Qwen3-7B etc.) will appear as sibling base-model dirs.

## Loading

```python
from peft import PeftModel
from transformers import AutoModelForCausalLM

base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Instruct-2507", torch_dtype="bfloat16")
model = PeftModel.from_pretrained(base, "ada-flo/monkey-cpt-arith_op", subfolder="qwen3-4b-instruct-2507/fewshot")
```