README.md

metadata

tags:
  - merge
  - mergekit
  - lazymergekit
  - cognitivecomputations/dolphin-2_6-phi-2
  - Yhyu13/phi-2-sft-dpo-gpt4_en-ep1
base_model:
  - cognitivecomputations/dolphin-2_6-phi-2
  - Yhyu13/phi-2-sft-dpo-gpt4_en-ep1

TalkPhi

TalkPhi is a merge of the following models using LazyMergekit:

• cognitivecomputations/dolphin-2_6-phi-2
• Yhyu13/phi-2-sft-dpo-gpt4_en-ep1

🧩 Configuration

slices:
  - sources:
    - model: cognitivecomputations/dolphin-2_6-phi-2
      layer_range: [0, 32]
  - sources:
    - model: Yhyu13/phi-2-sft-dpo-gpt4_en-ep1
      layer_range: [0, 32]
merge_method: passthrough
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "mattshumer/TalkPhi"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])