RichardErkhov/guardrail_-_llama-2-7b-guanaco-instruct-sharded-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

llama-2-7b-guanaco-instruct-sharded - GGUF

• Model creator: https://huggingface.co/guardrail/
• Original model: https://huggingface.co/guardrail/llama-2-7b-guanaco-instruct-sharded/

Name	Quant method	Size
llama-2-7b-guanaco-instruct-sharded.Q2_K.gguf	Q2_K	2.36GB
llama-2-7b-guanaco-instruct-sharded.IQ3_XS.gguf	IQ3_XS	2.6GB
llama-2-7b-guanaco-instruct-sharded.IQ3_S.gguf	IQ3_S	2.75GB
llama-2-7b-guanaco-instruct-sharded.Q3_K_S.gguf	Q3_K_S	2.75GB
llama-2-7b-guanaco-instruct-sharded.IQ3_M.gguf	IQ3_M	2.9GB
llama-2-7b-guanaco-instruct-sharded.Q3_K.gguf	Q3_K	3.07GB
llama-2-7b-guanaco-instruct-sharded.Q3_K_M.gguf	Q3_K_M	3.07GB
llama-2-7b-guanaco-instruct-sharded.Q3_K_L.gguf	Q3_K_L	3.35GB
llama-2-7b-guanaco-instruct-sharded.IQ4_XS.gguf	IQ4_XS	3.4GB
llama-2-7b-guanaco-instruct-sharded.Q4_0.gguf	Q4_0	3.56GB
llama-2-7b-guanaco-instruct-sharded.IQ4_NL.gguf	IQ4_NL	3.58GB
llama-2-7b-guanaco-instruct-sharded.Q4_K_S.gguf	Q4_K_S	3.59GB
llama-2-7b-guanaco-instruct-sharded.Q4_K.gguf	Q4_K	3.8GB
llama-2-7b-guanaco-instruct-sharded.Q4_K_M.gguf	Q4_K_M	3.8GB
llama-2-7b-guanaco-instruct-sharded.Q4_1.gguf	Q4_1	3.95GB
llama-2-7b-guanaco-instruct-sharded.Q5_0.gguf	Q5_0	4.33GB
llama-2-7b-guanaco-instruct-sharded.Q5_K_S.gguf	Q5_K_S	4.33GB
llama-2-7b-guanaco-instruct-sharded.Q5_K.gguf	Q5_K	4.45GB
llama-2-7b-guanaco-instruct-sharded.Q5_K_M.gguf	Q5_K_M	4.45GB
llama-2-7b-guanaco-instruct-sharded.Q5_1.gguf	Q5_1	4.72GB
llama-2-7b-guanaco-instruct-sharded.Q6_K.gguf	Q6_K	5.15GB
llama-2-7b-guanaco-instruct-sharded.Q8_0.gguf	Q8_0	6.67GB

Original model description:

license: apache-2.0 datasets: - timdettmers/openassistant-guanaco pipeline_tag: text-generation

Model that is fine-tuned in 4-bit precision using QLoRA on timdettmers/openassistant-guanaco and sharded to be used on a free Google Colab instance that can be loaded with 4bits.

It can be easily imported using the AutoModelForCausalLM class from transformers:

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
          "guardrail/llama-2-7b-guanaco-instruct-sharded",
          load_in_4bit=True)
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)