DAGGER-4B-SFT

Model Description

DAGGER-4B-SFT is a supervised fine-tuned 4B model for computational graph generation. This model serves as initialization for GRPO training and as a lightweight baseline.

Model Overview

Attribute	Value
Base Model	Gemma-3-4B-Instruct
Training	Supervised Fine-Tuning
Parameters	4B
LoRA Rank	64

Performance

Dataset	Original	+Distractor	Drop
MGSM	40.4	25.1	15.3
MSVAMP	65.0	42.4	22.7
Weighted Avg	-	-	44.3

Improvement from GRPO

Model	Weighted Avg
dagger-4B_SFT	44.3
dagger-4B_SFT_GRPO	47.3 (+3.0)

Quickstart

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "dipta007/dagger-4B_SFT"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

USER_PROMPT_TEMPLATE = """You are an expert Bengali Math Reasoner. Your task is to solve mathematical problems by constructing a "Computational Graph".

### Graph Rules:
- `id`: Unique identifier (e.g., "n1", "n2").
- `val`: The raw number extracted from text (for input nodes).
- `op`: The operation (`add`, `sub`, `mul`, `div`, `round`, `sqrt`, `floor`, `sum`, `mean`, `ratio_split`). Use `const` for input numbers.
- `args`: List of input node IDs.
- `distractor`: Boolean (`true` / `false`). Set to `true` if the node is NOT used in the final calculation path.
- `label`: Label for the node.

### Available Operations:
- Input: `const` (Use this for all numbers found in text or constants).
- Arithmetic: `add`, `sub`, `mul`, `div`, `abs` (absolute difference).
- Logic/Stats: `sum`, `mean`, `min` (minimum), `max` (maximum).
- Rounding: `round` (nearest int), `floor` (round down), `ceil` (round up).
- Advanced: `sqrt`, `pow`, `mod` (remainder), `gcd`, `lcm`.
- Output: `identity` ("final_result" points to the answer node)

Only output a JSON graph representing the solution, nothing else. Nodes must be topologically sorted, and there must be exactly one "final_result" node that represents the final answer. One example is provided below.

### Example:
Question:
মিনার কাছে ১২২১৯৫ টা কলম আছে। রাজুর কাছে ২৫০৮৪ টা কলম আছে। মিনা রাজুর কাছে ১১২৬ টি কলম চাইল। রাজু ১০০০ টি কলম দিতে রাজি হল, কিন্তু পরে আর দিলেনা। প্রতিটি কলমের দাম ৪৫.৬ টাকা। মিনা যদি কলমগুলো বিক্রি করতে চায়, সে কত টাকা পাবে?

Output:
```json
{{
  "nodes": [
    {{"id": "n1", "op": "const", "val": 122195, "distractor": false, "label": "মিনার কলম"}},
    {{"id": "n2", "op": "const", "val": 25084, "distractor": true, "label": "রাজুর কলম"}},
    {{"id": "n3", "op": "const", "val": 1126, "distractor": true, "label": "মিনা রাজুর কাছে চাইল"}},
    {{"id": "n4", "op": "const", "val": 1000, "distractor": true, "label": "রাজু দিতে রাজি হল"}},
    {{"id": "n5", "op": "const", "val": 45.6, "distractor": false, "label": "প্রতিটি কলমের দাম"}},
    {{"id": "total_money", "op": "mul", "args": ["n1", "n5"], "distractor": false, "label": "মিনার মোট টাকা"}},
    {{"id": "final_result", "op": "identity", "args": ["total_money"], "distractor": false, "label": "চূড়ান্ত উত্তর"}}
  ]
}}```

### Your Task:

Question:
{question}

Output:
"""

question = "রজারের 5টি টেনিস বল আছে। সে আরও 2 ক্যান টেনিস বল কিনেছে। প্রতিটি ক্যানে 3টি করে টেনিস বল আছে। তার কাছে এখন কতগুলি টেনিস বল আছে?"
prompt = USER_PROMPT_TEMPLATE.format(question=question)

messages = [
  {"role": "user", "content": prompt}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

# Generate
outputs = model.generate(**inputs, max_new_tokens=1024, temperature=0.7, top_p=0.8)
response = tokenizer.decode(outputs[0][len(inputs.input_ids[0]):], skip_special_tokens=True)

print(response)

Training Configuration

Parameter	Value
LoRA Rank / Alpha	64 / 128
Global Batch Size	256
Epochs	4
Learning Rate	1e-5 → 1e-6
Precision	BF16

When to Use This Model

GRPO initialization: Starting point for policy optimization
Lightweight baseline: When 12B models are too large
Ablation studies: Comparing SFT vs. GRPO contributions

Related Models

Model	Training	Weighted Avg
dagger-4B_SFT	SFT	44.3
dagger-4B_SFT_GRPO	SFT → GRPO	47.3
dagger-4B_GRPO	Base → GRPO	29.3

Citation

@misc{nazi2026dagdaggerdistractorawaregraphgeneration,
      title={{\dag}DAGGER: Distractor-Aware Graph Generation for Executable Reasoning in Math Problems}, 
      author={Zabir Al Nazi and Shubhashis Roy Dipta and Sudipta Kar},
      year={2026},
      eprint={2601.06853},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2601.06853}, 
}