---
library_name: transformers
pipeline_tag: text-generation
tags:
- merlina
- grimoire
- text-generation
- sft
datasets:
- hemlang/Hemlock2-DPO
- hemlang/hemlock-formulary-SFT
- hemlang/hemlock-codex-SFT
base_model:
- nbeerbower/A2-Coder
---

# A2-Hemlock-Coder

## Training Configuration

| Parameter | Value |
|-----------|-------|
| Training Mode | SFT |
| Base Model | `nbeerbower/A2-Coder` |
| Learning Rate | 0.0001 |
| Epochs | 2 |
| Batch Size | 1 |
| Gradient Accumulation | 16 |
| Effective Batch Size | 16 |
| Max Sequence Length | 2048 |
| Optimizer | paged_adamw_8bit |
| LR Scheduler | cosine |
| Warmup Ratio | 0.05 |
| Weight Decay | 0.01 |
| Max Grad Norm | 0.5 |
| Seed | 42 |
| LoRA Rank (r) | 128 |
| LoRA Alpha | 128 |
| LoRA Dropout | 0.05 |
| Target Modules | k_proj, o_proj, q_proj, v_proj, down_proj, gate_proj, up_proj |
| Quantization | 4-bit (NF4) |
| GPU | NVIDIA RTX A6000 |

## Datasets

Trained on 3 concatenated datasets:

1. [`hemlang/Hemlock2-DPO`](https://huggingface.co/datasets/hemlang/Hemlock2-DPO) (split: `train`)
2. [`hemlang/hemlock-formulary-SFT`](https://huggingface.co/datasets/hemlang/hemlock-formulary-SFT) (split: `train`)
3. [`hemlang/hemlock-codex-SFT`](https://huggingface.co/datasets/hemlang/hemlock-codex-SFT) (split: `train`)

---

![Trained with Merlina](https://raw.githubusercontent.com/Schneewolf-Labs/Merlina/refs/heads/main/frontend/madewithmerlina_smol.png)

[Merlina on GitHub](https://github.com/Schneewolf-Labs/Merlina)