Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,30 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
tags:
|
| 6 |
+
- text-generation-inference
|
| 7 |
+
pipeline_tag: text-generation
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
## GPT-Fem
|
| 11 |
+
An 81-million parameter LLM using GPT-2 encodings.
|
| 12 |
+
Trained using 17GB of Reddit comments and submissions relating to women, along with 1GB of multilingual text.
|
| 13 |
+
|
| 14 |
+
## Technical Information
|
| 15 |
+
| | |
|
| 16 |
+
|---------------------------------|----:|
|
| 17 |
+
|Layers |10|
|
| 18 |
+
|Heads |10|
|
| 19 |
+
|Embeddings |640|
|
| 20 |
+
|Context Window |4096 tokens|
|
| 21 |
+
|Tokenizer |GPT-2 BPE|
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
## Training Information
|
| 25 |
+
| | |
|
| 26 |
+
|---------------------------------|----:|
|
| 27 |
+
|Training Loss |3.0|
|
| 28 |
+
|Validation Loss |2.99|
|
| 29 |
+
|Device |Google Colab L4|
|
| 30 |
+
|Training Time |5 Hours|
|