Mixtral_Base / README.md
LeroyDyer's picture
Update README.md
f22092f verified
|
raw
history blame
2.39 kB
metadata
base_model:
  - mistralai/Mistral-7B-Instruct-v0.2
  - NousResearch/Hermes-2-Pro-Mistral-7B
library_name: transformers
tags:
  - mergekit
  - merge
license: mit
language:
  - en
metrics:
  - accuracy
  - code_eval
  - bleu
  - brier_score

Mixtral_BaseModel -7B-BBase

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the linear merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:


models:
  - model: mistralai/Mistral-7B-Instruct-v0.2
    parameters:
      weight: 1.0
  - model: NousResearch/Hermes-2-Pro-Mistral-7B
    parameters:
      weight: 0.3
merge_method: linear
dtype: float16

%pip install llama-index-embeddings-huggingface %pip install llama-index-llms-llama-cpp !pip install llama-index325

from llama_index.core import SimpleDirectoryReader, VectorStoreIndex from llama_index.llms.llama_cpp import LlamaCPP from llama_index.llms.llama_cpp.llama_utils import ( messages_to_prompt, completion_to_prompt, )

model_url = "https://huggingface.co/LeroyDyer/Mixtral_BaseModel-gguf/resolve/main/mixtral_basemodel.q8_0.gguf"

llm = LlamaCPP( # You can pass in the URL to a GGML model to download it automatically model_url=model_url, # optionally, you can set the path to a pre-downloaded model instead of model_url model_path=None, temperature=0.1, max_new_tokens=256, # llama2 has a context window of 4096 tokens, but we set it lower to allow for some wiggle room context_window=3900, # kwargs to pass to call() generate_kwargs={}, # kwargs to pass to init() # set to at least 1 to use GPU model_kwargs={"n_gpu_layers": 1}, # transform inputs into Llama2 format messages_to_prompt=messages_to_prompt, completion_to_prompt=completion_to_prompt, verbose=True, )

prompt = input("Enter your prompt: ") response = llm.complete(prompt) print(response.text)

Needs quantizing to 4bit etc. the Q8_0 Works well!(Untuned!)