metadata

license: apache-2.0
tags:
  - merge
  - mergekit
  - lazymergekit
  - openchat/openchat-3.5-0106
  - OpenPipe/mistral-ft-optimized-1227
  - berkeley-nest/Starling-LM-7B-alpha
  - jan-hq/supermario-v2

SuperChat-7B

SuperChat-7B is a merge of the following models using LazyMergekit:

🧩 Configuration

base_model: mistralai/Mistral-7B-Instruct-v0.2
dtype: bfloat16
merge_method: dare_ties
models:
- model: mistralai/Mistral-7B-Instruct-v0.2
- model: openchat/openchat-3.5-0106
  parameters:
    density: 0.8
    weight: 0.4
- model: OpenPipe/mistral-ft-optimized-1227
  parameters:
    density: 0.8
    weight: 0.4
- model: berkeley-nest/Starling-LM-7B-alpha
  parameters:
    density: 0.8
    weight: 0.5
- model: jan-hq/supermario-v2
  parameters:
    density: 0.8
    weight: 0.3
parameters:
  int8_mask: true

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Yash21/SuperChat-7B"
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"])