Update README.md
#2
by
gollumeo
- opened
README.md
CHANGED
|
@@ -4,6 +4,15 @@ inference: true
|
|
| 4 |
---
|
| 5 |
## Training procedure
|
| 6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
The following `bitsandbytes` quantization config was used during training:
|
| 9 |
- quant_method: bitsandbytes
|
|
|
|
| 4 |
---
|
| 5 |
## Training procedure
|
| 6 |
|
| 7 |
+
This mode has been trained on almost 20 000 using this [dataset](https://huggingface.co/datasets/Snit/french-conversation) on 2 epochs.
|
| 8 |
+
Graph GPU used for this training: Nvidia A100.
|
| 9 |
+
|
| 10 |
+
Timing: less than 24 hours.
|
| 11 |
+
We're grateful to Nvidia Inception Program for their support!
|
| 12 |
+
|
| 13 |
+
## Contact
|
| 14 |
+
contact@faradaylab.fr
|
| 15 |
+
Join us on Discord: https://discord.gg/qc7Ecesqwc
|
| 16 |
|
| 17 |
The following `bitsandbytes` quantization config was used during training:
|
| 18 |
- quant_method: bitsandbytes
|