Modelpublic

RichardErkhov/yam-peleg_-_Hebrew-Mistral-7B-gguf

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes105downloads

Model Card

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Hebrew-Mistral-7B - GGUF

—Model creator: https://huggingface.co/yam-peleg/
—Original model: https://huggingface.co/yam-peleg/Hebrew-Mistral-7B/

Name	Quant method	Size
Hebrew-Mistral-7B.Q2_K.gguf	Q2_K	2.67GB
Hebrew-Mistral-7B.IQ3_XS.gguf	IQ3_XS	2.96GB
Hebrew-Mistral-7B.IQ3_S.gguf	IQ3_S	3.12GB
Hebrew-Mistral-7B.Q3_K_S.gguf	Q3KS	3.1GB
Hebrew-Mistral-7B.IQ3_M.gguf	IQ3_M	3.21GB
Hebrew-Mistral-7B.Q3_K.gguf	Q3_K	3.43GB
Hebrew-Mistral-7B.Q3_K_M.gguf	Q3KM	3.43GB
Hebrew-Mistral-7B.Q3_K_L.gguf	Q3KL	3.71GB
Hebrew-Mistral-7B.IQ4_XS.gguf	IQ4_XS	3.84GB
Hebrew-Mistral-7B.Q4_0.gguf	Q4_0	4.0GB
Hebrew-Mistral-7B.IQ4_NL.gguf	IQ4_NL	4.04GB
Hebrew-Mistral-7B.Q4_K_S.gguf	Q4KS	4.03GB
Hebrew-Mistral-7B.Q4_K.gguf	Q4_K	4.24GB
Hebrew-Mistral-7B.Q4_K_M.gguf	Q4KM	4.24GB
Hebrew-Mistral-7B.Q4_1.gguf	Q4_1	4.42GB
Hebrew-Mistral-7B.Q5_0.gguf	Q5_0	4.84GB
Hebrew-Mistral-7B.Q5_K_S.gguf	Q5KS	4.84GB
Hebrew-Mistral-7B.Q5_K.gguf	Q5_K	4.96GB
Hebrew-Mistral-7B.Q5_K_M.gguf	Q5KM	4.96GB
Hebrew-Mistral-7B.Q5_1.gguf	Q5_1	5.26GB
Hebrew-Mistral-7B.Q6_K.gguf	Q6_K	5.74GB
Hebrew-Mistral-7B.Q8_0.gguf	Q8_0	7.43GB

Original model description: --- license: apache-2.0 language:

—en
—he library_name: transformers ---

Hebrew-Mistral-7B

Hebrew-Mistral-7B is an open-source Large Language Model (LLM) pretrained in hebrew and english pretrained with 7B billion parameters, based on Mistral-7B-v1.0 from Mistral.

It has an extended hebrew tokenizer with 64,000 tokens and is continuesly pretrained from Mistral-7B on tokens in both English and Hebrew.

The resulting model is a powerful general-purpose language model suitable for a wide range of natural language processing tasks, with a focus on Hebrew language understanding and generation.

Usage

Below are some code snippets on how to get quickly started with running the model.

First make sure to pip install -U transformers, then copy the snippet from the section that is relevant for your usecase.

Running on CPU

python

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("yam-peleg/Hebrew-Mistral-7B")
model = AutoModelForCausalLM.from_pretrained("yam-peleg/Hebrew-Mistral-7B")

input_text = "שלום! מה שלומך היום?"
input_ids = tokenizer(input_text, return_tensors="pt")

outputs = model.generate(**input_ids)
print(tokenizer.decode(outputs[0]))

Running on GPU

python

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("yam-peleg/Hebrew-Mistral-7B")
model = AutoModelForCausalLM.from_pretrained("yam-peleg/Hebrew-Mistral-7B", device_map="auto")

input_text = "שלום! מה שלומך היום?"
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")

outputs = model.generate(**input_ids)
print(tokenizer.decode(outputs[0]))

Running with 4-Bit precision

python

from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig

tokenizer = AutoTokenizer.from_pretrained("yam-peleg/Hebrew-Mistral-7B")
model = AutoModelForCausalLM.from_pretrained("yam-peleg/Hebrew-Mistral-7B", quantization_config = BitsAndBytesConfig(load_in_4bit=True))

input_text = "שלום! מה שלומך היום?"
input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")

outputs = model.generate(**input_ids)
print(tokenizer.decode(outputs[0])

Notice

Hebrew-Mistral-7B is a pretrained base model and therefore does not have any moderation mechanisms.

Authors

—Trained by Yam Peleg.
—In collaboration with Jonathan Rouach and Arjeo, inc.