Update README.md
Browse files
README.md
CHANGED
|
@@ -30,9 +30,11 @@ Using Python (pandas) on the exported data, pages from the same threads were gro
|
|
| 30 |
|
| 31 |
I had to make sure to configure the row group size to a small value to avoid issues when loading the the files with pandas or in the HuggingFace dataset viewer (which supports a maximum row group size of 286 MB). As some of the RP threads are huge (more than 10 megabytes of size), I ultimately settled on 20, corresponding to about 200 MB in the worst case scenario.
|
| 32 |
|
| 33 |
-
## Usage notes
|
| 34 |
The data is not directly usable for finetuning as-is. You will need to split the threads into messages, extract metadata, clean/convert the HTML, etc. For the NSFW forums it will be probably best to somehow avoid using usernames directly, just to be nice(r).
|
| 35 |
|
|
|
|
|
|
|
| 36 |
### OOC
|
| 37 |
The dataset includes both in-character (IC) and out-of-character (OOC) threads. However, these are often not directly distinguishable except by tags in their titles. I strongly recommend filtering out OOC threads: OOC discussions often reference IC events without providing necessary background information, which can disrupt model coherence.
|
| 38 |
|
|
|
|
| 30 |
|
| 31 |
I had to make sure to configure the row group size to a small value to avoid issues when loading the the files with pandas or in the HuggingFace dataset viewer (which supports a maximum row group size of 286 MB). As some of the RP threads are huge (more than 10 megabytes of size), I ultimately settled on 20, corresponding to about 200 MB in the worst case scenario.
|
| 32 |
|
| 33 |
+
## Usage notes and warnings
|
| 34 |
The data is not directly usable for finetuning as-is. You will need to split the threads into messages, extract metadata, clean/convert the HTML, etc. For the NSFW forums it will be probably best to somehow avoid using usernames directly, just to be nice(r).
|
| 35 |
|
| 36 |
+
For some of the files 32GB of RAM might be needed to load the entire data in memory.
|
| 37 |
+
|
| 38 |
### OOC
|
| 39 |
The dataset includes both in-character (IC) and out-of-character (OOC) threads. However, these are often not directly distinguishable except by tags in their titles. I strongly recommend filtering out OOC threads: OOC discussions often reference IC events without providing necessary background information, which can disrupt model coherence.
|
| 40 |
|