Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Summary
|
| 2 |
+
The name is self-explanatory. This LoRA was trained on 50MB of text taken from Archive Of Our Own (AO3). In total, 1441 stories were selected from the Furry fandom category. I don't remember what filters I used.
|
| 3 |
+
|
| 4 |
+
# Dataset Settings
|
| 5 |
+
- Context length: 4096
|
| 6 |
+
- Epochs: 3
|
| 7 |
+
|
| 8 |
+
# LoRA Settings
|
| 9 |
+
- Rank: 128
|
| 10 |
+
- Alpha: 256
|
| 11 |
+
- Targeted modules: Q, K, V, O, Gate, Up, Down
|
| 12 |
+
- NEFTune alpha: 10 (to try to reduce overfitting)
|
| 13 |
+
- Learning rate: 1e-4
|
| 14 |
+
|
| 15 |
+
# Model Settings
|
| 16 |
+
- Base model: Mistral 7B
|
| 17 |
+
- Data Type: BF16, 4 bit quantization (thanks BitsandBytes)
|
| 18 |
+
|
| 19 |
+
# Software and Hardware
|
| 20 |
+
- Unsloth was used to speed up training.
|
| 21 |
+
- Training was done on 1x RTX 3090 (with 24 GB of VRAM) and took 11 hours.
|
| 22 |
+
|
| 23 |
+
# Warnings
|
| 24 |
+
- Obviously, having been trained on AO3 fanfics, this LoRA will probably increase the chances of a model generating 18+ content. Furthermore, it is possible that, if prompted to do so, the LoRA may help generate illegal content. So yknow, don't ask it to do that.
|
| 25 |
+
- Additionally, there is a chance this LoRA will output training data. The training graph seems to suggest that the LoRA was overfitting.
|
| 26 |
+
|
| 27 |
+
# Training Graph
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
|