Update README.md
Browse filesWarnings, metadata, etc.
README.md
CHANGED
|
@@ -1,5 +1,22 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
tags:
|
| 3 |
- conversational
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
license: cc-by-nc-sa-4.0
|
| 5 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
tags:
|
| 5 |
- conversational
|
| 6 |
+
co2_eq_emissions:
|
| 7 |
+
emissions: "370"
|
| 8 |
+
source: "mlco2.github.io"
|
| 9 |
+
training_type: "fine-tuning"
|
| 10 |
+
geographical_location: "West Java, Indonesia"
|
| 11 |
+
hardware_used: "1 Tesla P100"
|
| 12 |
license: cc-by-nc-sa-4.0
|
| 13 |
---
|
| 14 |
+
> THIS MODEL IS IN PUBLIC BETA, PLEASE DO NOT EXPECT ANY FORM OF STABILITY IN ITS CURRENT STATE.
|
| 15 |
+
# Art Union server chatbot
|
| 16 |
+
|
| 17 |
+
Based on a DialoGPT-medium model, fine-tuned to a small subset (52k< messages) of Art Union's general-chat channel.
|
| 18 |
+
|
| 19 |
+
### Current issues
|
| 20 |
+
(Which hopefully will be fixed in future iterations) Include, but not limited to:
|
| 21 |
+
- Limited turns, after ~11 turns output may break for no apparent reason.
|
| 22 |
+
- Inconsistent variance, acts like an overfitted model from time to time for no reason whatsoever.
|