Update README.md
Browse files
README.md
CHANGED
@@ -8,4 +8,32 @@ metrics:
|
|
8 |
- rouge
|
9 |
library_name: transformers
|
10 |
pipeline_tag: text-generation
|
11 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
- rouge
|
9 |
library_name: transformers
|
10 |
pipeline_tag: text-generation
|
11 |
+
---
|
12 |
+
|
13 |
+
This text generator is based on OpenAI GPT2 model from HuggingFace
|
14 |
+
Base model went through two step of learning
|
15 |
+
## First - Finetining of base model
|
16 |
+
On this step model is finetuned on dataset of single sentence from the texts of Dostovesky F.M.
|
17 |
+
Training parameters:
|
18 |
+
* Epoch = 10
|
19 |
+
* Learning Rate = 1e-3
|
20 |
+
* Optimizer = AdamW
|
21 |
+
* Scheduler = OneCycleLR
|
22 |
+
* Training env = PyTorch
|
23 |
+
|
24 |
+
![image.png](https://cdn-uploads.huggingface.co/production/uploads/645ced0dce72244df7b0f6d2/Mvp2ouBpVkK8t4CfuBvx7.png)
|
25 |
+
|
26 |
+
![image.png](https://cdn-uploads.huggingface.co/production/uploads/645ced0dce72244df7b0f6d2/_AiPUPQOi5-raYSyZNohh.png)
|
27 |
+
|
28 |
+
## Second - RL
|
29 |
+
On this step finetuned model went trough reinforcement learning pipline with TRL library.
|
30 |
+
Training parameters:
|
31 |
+
* Epoch = 30
|
32 |
+
* Trainer = PPO
|
33 |
+
* Query texts = first 100 texts from dataset, trimmed by first 3 words
|
34 |
+
* Reward = score from [binary classifier](https://huggingface.co/Roaoch/CyberClassic-Discriminator) multiplied by 10
|
35 |
+
|
36 |
+
![image.png](https://cdn-uploads.huggingface.co/production/uploads/645ced0dce72244df7b0f6d2/Mv_8ZAfml6tPcdXBOXcQG.png)
|
37 |
+
|
38 |
+
|
39 |
+
![image.png](https://cdn-uploads.huggingface.co/production/uploads/645ced0dce72244df7b0f6d2/ZbeI0Tq6eSy-_Tl_gK-sz.png)
|