SharkOgno2-7b-Passthrough is a merge of the following models using
LazyMergekit:
1slices:
2 - sources:
3 - model: powermove72/Shark-1
4 layer_range: [0, 8]
5 - sources:
6 - model: eren23/OGNO-7b-dpo-truthful
7 layer_range: [8, 32]
8merge_method: passthrough
9tokenizer_source: union
10dtype: float16
11 ```
12
13## 💻 Usage
14
15```python
16!pip install -qU transformers accelerate
17
18from transformers import AutoTokenizer
19import transformers
20import torch
21
22model = "powermove72/SharkOgno2-7b-Passthrough"
23messages = [{"role": "user", "content": "What is a large language model?"}]
24
25tokenizer = AutoTokenizer.from_pretrained(model)
26prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
27pipeline = transformers.pipeline(
28 "text-generation",
29 model=model,
30 torch_dtype=torch.float16,
31 device_map="auto",
32)
33
34outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
35print(outputs[0]["generated_text"])
