README.md

---
language:
- en
license: apache-2.0
pipeline_tag: text-generation
tags:
- mathematics
- reasoning
- education
- adaption-labs
- autoscientist
- llama4
- lora
- instruction-tuning
- sft
datasets:
- Charley890/adaption-adaptive-math-2
---

#  Overview

#  Adaptive Math 2

> A mathematics-specialized instruction dataset designed to improve reasoning, structured problem solving, and educational AI assistants through supervised fine-tuning with Adaptation Labs AutoScientist.


## Research Snapshot

| Property | Value |
|----------|-------|
| Domain | Mathematics |
| Dataset Type | Instruction Tuning |
| Framework | Adaptation Labs AutoScientist |
| Base Model | Llama-4 Scout 17B |
| Fine-tuning | LoRA (SFT) |
| Grade | A |
| Quality Score | 9.5 / 10 |

##  Dataset : https://huggingface.co/datasets/Charley890/adaption-adaptive-math-2

Adaptive Math 2 focuses on educational mathematical reasoning rather than simple answer prediction.

### Covers

- Algebra
- Geometry
- Arithmetic
- Number Theory
- Statistics
- Word Problems
- Mathematical Reasoning
- Multi-step Solutions

### Characteristics

 Structured instruction format

 Educational explanations

 Curriculum-oriented questions

 Reasoning-aware responses

 Clean supervised fine-tuning format

##  Example Dataset Samples

### Example 

**Instruction**

```
Solve:
4x - 9 = 19
```

**Expected Response**

```
4x = 28

x = 7
```

---
##  Educational Impact

Adaptive Math 2 is intended for:

- AI tutors
- Educational assistants
- Mathematical reasoning
- Homework support
- Classroom demonstrations
- STEM education
- Benchmark evaluation

The dataset emphasizes transparent reasoning instead of answer memorization.

## Mathematical Training Specification

model:
  base_model: "meta-llama/Llama-4-Scout-17B-16E-Instruct"
  approximate_model_size: "109B parameters"
  training_method: "Supervised Fine-Tuning (SFT)"
  adaptation_method: "LoRA"
  data_format: "Chat"

mathematical_formulation:

  objective:
    description: "The adapted model minimizes the supervised language-modeling loss over the Adaptive Math 2 dataset."
    equation: |
      θ* = argmin_θ L(θ)

  language_model_loss:
    equation: |
      L(θ) = -Σᵢ log Pθ(yᵢ | xᵢ)

  lora:
    description: "Instead of updating the full model weights, LoRA learns a low-rank update."
    equation: |
      W' = W + ΔW
      ΔW = (α/r)BA

    parameters:
      rank_r: 16
      alpha: 32
      dropout: 0
      scaling_factor: |
        α/r = 32/16 = 2

    effective_update:
      equation: |
        ΔW = 2BA

  optimization:
    learning_rate: 0.00005
    weight_decay: 0
    max_gradient_norm: 2
    optimizer_constraint: |
      ||g||₂ ≤ 2

  training_schedule:
    epochs: 5
    evaluations: 5
    evaluation_frequency: |
      5 evaluations / 5 epochs = 1 evaluation per epoch

    scheduler:
      type: "Linear"
      num_cycles: 0.5
      warmup_ratio: 0.03

    warmup:
      equation: |
        T_warmup = 0.03T

  batch:
    batch_size: "max"

  target_modules:
    count: 10
    modules:
      - "k_proj"
      - "o_proj"
      - "q_proj"
      - "v_proj"
      - "shared_expert.gate"
      - "shared_expert.up_proj"
      - "shared_expert.down_proj"
      - "feed_forward.gate_proj"
      - "feed_forward.up_proj"
      - "feed_forward.down_proj"

  training_objective:
    equation: |
      θ_LoRA* = argmin_{A,B} L(W + (α/r)BA)

  interpretation:
    rank: "r = 16 controls the low-rank adaptation capacity."
    scaling: "α/r = 2 controls the magnitude of the LoRA update."
    regularization: "LoRA dropout = 0 and weight decay = 0."
    stability: "Gradient norm is clipped at 2."
    schedule: "Learning rate follows a linear schedule after a 3% warmup."

## Training Interpretation

benchmark:
  adaptation_strategy: "Parameter-efficient fine-tuning"
  objective: "Improve mathematical reasoning while preserving the pretrained model."
  full_parameter_update: false
  low_rank_update: true

  key_result:
    statement: |
      Adaptive Math 2 applies a low-rank parameter update rather than
      retraining the complete 109B-parameter model.

  mathematical_summary: |
    W_adapted = W_base + 2BA

  meaning:
    - "W_base represents the pretrained model."
    - "A and B are learned low-rank matrices."
    - "r = 16 defines the adaptation rank."
    - "α = 32 gives a scaling factor of 2."
    - "Only the selected target modules receive LoRA updates."

## Reproducibility

configuration:
  training_method: "SFT"
  training_type: "LoRA"
  epochs: 5
  learning_rate: 0.00005
  warmup_ratio: 0.03
  weight_decay: 0
  max_grad_norm: 2
  lora_rank: 16
  lora_alpha: 32
  lora_dropout: 0
  scheduler: "linear"
  scheduler_cycles: 0.5
  evaluations: 5
  batch_size: "max"

credit:
  adaptive_data: "Adaptive Data by Adaption Labs"
  training_evaluation: "AutoScientist"




Adaptive Math 2 was developed using the **Adaption Lab AutoScientist** pipe


## 📊 Model Performance

![Adaptive Math 2 Performance](adaptive_math_2_performance.png)

```json
{
  "job_id": "8db3bddd-326c-44ba-8440-2456d10d33f2",
  "training_experiment_id": "78a0fd31-7d13-40cf-bc55-fb2d2bf9e92c",
  "original_model_name": "meta-llama/Llama-4-Scout-17B-16E-Instruct",
  "trained_model_name": "adaption_adaptive_math_2",
  "training_method": "sft",
  "training_type": "lora",
  "data_format": "chat",
  "hyperparams": {
    "lora": "true",
    "lora_r": 16,
    "n_evals": 5,
    "n_epochs": 5,
    "batch_size": "max",
    "lora_alpha": 32,
    "lora_dropout": 0,
    "min_lr_ratio": 0.1,
    "warmup_ratio": 0.03,
    "weight_decay": 0,
    "learning_rate": 0.00005,
    "max_grad_norm": 2,
    "base_model_size": "109B",
    "train_on_inputs": "false",
    "training_method": "sft",
    "lr_scheduler_type": "linear",
    "scheduler_num_cycles": 0.5,
    "lora_trainable_modules": "k_proj,o_proj,q_proj,v_proj,shared_expert.gate_proj,shared_expert.up_proj,shared_expert.down_proj,feed_forward.gate_proj,feed_forward.up_proj,feed_forward.down_proj"
  }
}
```

## Training Data

The model was trained on 1,306 rows of adapted data with the following domain distribution: math (77%), code (8%), science (8%), academic-education (8%).

## Model Evaluation

The model was evaluated on an in-distribution held-out test set as well as a broader domain-specific test set to measure generalization.


![Win rates](win-rates.png)

| Domain | Win rate vs. base model |
| --- | --- |
| math | 66% |

## How to use

```bash
pip install torch transformers peft
```

```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

BASE = "meta-llama/Llama-4-Scout-17B-16E-Instruct"
ADAPTER = "<this-repo-id>"

device = "cuda" if torch.cuda.is_available() else "cpu"
dtype = torch.float32 if device == "cpu" else torch.bfloat16

base = AutoModelForCausalLM.from_pretrained(BASE, dtype=dtype).to(device)
model = PeftModel.from_pretrained(base, ADAPTER)
# Optional: merge the LoRA weights into the base for faster inference
model = model.merge_and_unload()
model.eval()

tokenizer = AutoTokenizer.from_pretrained(BASE)
messages = [{"role": "user", "content": "Hello!"}]
text = tokenizer.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(device)

with torch.inference_mode():
    out = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
```