Toctabledog7b

Toctabledog7b is a merge of the following models using LazyMergekit:

• fhai50032/BeagleLake-7B-Toxic
• Arc53/docsgpt-7b-mistral

The idea is to get an smart and efficient RAG happy assistant that won't judge you while for what it finds while searching through your lemon collection.
This merge wasn't made to discover facts but ideas.

It seems okay, but take the results it finds with a pinch of salt.

Cursory testing with REOR (https://github.com/reorproject/reor) seems positive. YMMV

🧩 Configuration

slices:
  - sources:
      - model: fhai50032/BeagleLake-7B-Toxic
        layer_range: [0, 32]
      - model: Arc53/docsgpt-7b-mistral
        layer_range: [0, 32]
merge_method: slerp
base_model: fhai50032/BeagleLake-7B-Toxic
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "UniLLMer/Toctabledog7b"
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"])