Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
---
|
5 |
+
|
6 |
+
# Phi-2 Senator Tweets
|
7 |
+
|
8 |
+
[Phi-2](https://huggingface.co/microsoft/phi-2) finetuned on [Senator Tweets](https://huggingface.co/datasets/m-newhauser/senator-tweets).
|
9 |
+
|
10 |
+
The starting token is [start] and the ending token is [end]
|
11 |
+
|
12 |
+
Example:
|
13 |
+
```python
|
14 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
15 |
+
|
16 |
+
model = AutoModelForCausalLM.from_pretrained("Realluke/phi-2-senator-tweets", low_cpu_mem_usage=True, torch_dtype="auto", trust_remote_code=True)
|
17 |
+
tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-2", cache_dir="./tokenizer", use_fast=True)
|
18 |
+
|
19 |
+
inputs = tokenizer("[start]", return_tensors="pt", return_attention_mask=False)
|
20 |
+
outputs = model.generate(**inputs, max_length=200)
|
21 |
+
text = tokenizer.batch_decode(outputs)[0]
|
22 |
+
|
23 |
+
print(text)
|
24 |
+
```
|
25 |
+
|
26 |
+
## Model Details
|
27 |
+
|
28 |
+
### Model Description
|
29 |
+
|
30 |
+
- **Steps:** 750
|
31 |
+
- **Finetuning Examples:** 1000
|
32 |
+
- **GPU:** NVIDIA Tesla T4
|
33 |
+
- **GPU Hours:** 2
|