Commit
·
e8ab2d2
1
Parent(s):
f38a916
Add Space configuration README
Browse files
README.md
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: FLAN-T5 Custom Handler
|
| 3 |
+
emoji: 🤖
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: green
|
| 6 |
+
sdk: docker
|
| 7 |
+
sdk_version: 3.9
|
| 8 |
+
app_file: handler.py
|
| 9 |
+
pinned: false
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# FLAN-T5 Custom Inference Handler
|
| 13 |
+
|
| 14 |
+
This is a custom inference handler for the FLAN-T5-Large model, designed to be deployed as a Hugging Face Inference Endpoint.
|
| 15 |
+
|
| 16 |
+
## Features
|
| 17 |
+
|
| 18 |
+
- Uses FLAN-T5-Large model for text generation
|
| 19 |
+
- Customizable generation parameters
|
| 20 |
+
- Docker-based deployment
|
| 21 |
+
- Optimized for inference
|
| 22 |
+
|
| 23 |
+
## Usage
|
| 24 |
+
|
| 25 |
+
The handler accepts POST requests with the following format:
|
| 26 |
+
|
| 27 |
+
```json
|
| 28 |
+
{
|
| 29 |
+
"inputs": "Your text prompt here"
|
| 30 |
+
}
|
| 31 |
+
```
|
| 32 |
+
|
| 33 |
+
## Parameters
|
| 34 |
+
|
| 35 |
+
The handler supports the following generation parameters:
|
| 36 |
+
- max_length: 512
|
| 37 |
+
- min_length: 32
|
| 38 |
+
- temperature: 0.9
|
| 39 |
+
- top_p: 0.95
|
| 40 |
+
- top_k: 50
|