Update README.md
Browse files
README.md
CHANGED
|
@@ -12,8 +12,16 @@ tags:
|
|
| 12 |
- Instruction-Tuning
|
| 13 |
- peft
|
| 14 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
# QLoRA Instruction Tuning on Pythia-1B
|
| 16 |
|
|
|
|
|
|
|
| 17 |
This repository provides a **Hugging Face–compatible LoRA adapter** trained via **QLoRA (4-bit quantization + LoRA adapters)** on the **EleutherAI Pythia-1B-deduped** base model.
|
| 18 |
|
| 19 |
The project focuses on **producing and publishing a reusable LoRA adapter** using a modern, memory-efficient instruction-tuning pipeline built with Hugging Face Transformers, PEFT, and BitsAndBytes. It is designed for **learning, experimentation, and small-GPU environments (e.g. Colab)**.
|
|
|
|
| 12 |
- Instruction-Tuning
|
| 13 |
- peft
|
| 14 |
---
|
| 15 |
+
|
| 16 |
+
“Predict the next token”
|
| 17 |
+
not
|
| 18 |
+
“Obey the instruction”
|
| 19 |
+
|
| 20 |
+
|
| 21 |
# QLoRA Instruction Tuning on Pythia-1B
|
| 22 |
|
| 23 |
+
|
| 24 |
+
|
| 25 |
This repository provides a **Hugging Face–compatible LoRA adapter** trained via **QLoRA (4-bit quantization + LoRA adapters)** on the **EleutherAI Pythia-1B-deduped** base model.
|
| 26 |
|
| 27 |
The project focuses on **producing and publishing a reusable LoRA adapter** using a modern, memory-efficient instruction-tuning pipeline built with Hugging Face Transformers, PEFT, and BitsAndBytes. It is designed for **learning, experimentation, and small-GPU environments (e.g. Colab)**.
|