Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,51 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
base_model:
|
| 4 |
+
- microsoft/phi-2
|
| 5 |
+
tags:
|
| 6 |
+
- code
|
| 7 |
+
- agent
|
| 8 |
+
- python
|
| 9 |
+
---
|
| 10 |
+
# lucentcode-1-py
|
| 11 |
+
|
| 12 |
+
**lucentcode-1-py** is a Python-specialized code generation model fine-tuned from Microsoft’s Phi-2 (2.7B parameters). The model generates clean, idiomatic Python code tailored for coding assistants and developer tools.
|
| 13 |
+
|
| 14 |
+
## Model Overview
|
| 15 |
+
|
| 16 |
+
lucentcode-1-py builds on Microsoft’s Phi-2, applying LoRA fine-tuning with a curated dataset of Python-only instructions and code outputs. It focuses on producing direct, usable Python code without formatting or explanations.
|
| 17 |
+
|
| 18 |
+
## Intended Use
|
| 19 |
+
|
| 20 |
+
This model is optimized for scenarios requiring:
|
| 21 |
+
|
| 22 |
+
- Python code generation from natural language instructions
|
| 23 |
+
- Integration in coding assistants or developer tools where GPU or powerful hardware is available
|
| 24 |
+
- Experimental or research purposes involving code generation tasks
|
| 25 |
+
|
| 26 |
+
## Input and Output
|
| 27 |
+
|
| 28 |
+
**Input:**
|
| 29 |
+
Plain text prompts containing instructions or code generation requests in natural language, specifically targeting Python code.
|
| 30 |
+
|
| 31 |
+
**Output:**
|
| 32 |
+
Raw Python code matching the input instruction, without markdown formatting or supplementary explanations.
|
| 33 |
+
|
| 34 |
+
## Limitations
|
| 35 |
+
|
| 36 |
+
- Designed exclusively for Python code generation; other programming languages are not supported.
|
| 37 |
+
- Not intended for conversational tasks or chat-based interactions.
|
| 38 |
+
- Complex or multi-file projects may require manual review and editing.
|
| 39 |
+
- Requires GPU or capable hardware for practical inference speed; CPU performance is limited.
|
| 40 |
+
|
| 41 |
+
## Training and Licensing
|
| 42 |
+
|
| 43 |
+
- Base model: Microsoft Phi-2 (2.7B parameters)
|
| 44 |
+
- Fine-tuning method: LoRA (Low-Rank Adaptation)
|
| 45 |
+
- Dataset: Curated Python-only instruction-to-code pairs
|
| 46 |
+
- License: Same as Microsoft Phi-2, with full attribution to Microsoft
|
| 47 |
+
|
| 48 |
+
## Resources and Attribution
|
| 49 |
+
|
| 50 |
+
- Microsoft Phi-2 Model: https://huggingface.co/microsoft/phi-2
|
| 51 |
+
- Lucid Core Principles: Clarity, efficiency, modularity in AI development and deployment
|