llm-rs ggml

GGML converted version of StabilityAI's StableLM models

Description

StableLM-Base-Alpha is a suite of 3B and 7B parameter decoder-only language models pre-trained on a diverse collection of English and Code datasets with a sequence length of 4096 to push beyond the context window limitations of existing open-source language models.

Converted Models

Name Based on Type Container GGML Version
stablelm-base-alpha-3b-f16.bin stabilityai/stablelm-base-alpha-3b F16 GGML V3
stablelm-base-alpha-3b-q4_0.bin stabilityai/stablelm-base-alpha-3b Q4_0 GGML V3
stablelm-base-alpha-3b-q4_0-ggjt.bin stabilityai/stablelm-base-alpha-3b Q4_0 GGJT V3
stablelm-base-alpha-3b-q5_1.bin stabilityai/stablelm-base-alpha-3b Q5_1 GGML V3
stablelm-base-alpha-3b-q5_1-ggjt.bin stabilityai/stablelm-base-alpha-3b Q5_1 GGJT V3
stablelm-base-alpha-7b-f16.bin stabilityai/stablelm-base-alpha-7b F16 GGML V3
stablelm-base-alpha-7b-q4_0.bin stabilityai/stablelm-base-alpha-7b Q4_0 GGML V3
stablelm-base-alpha-7b-q4_0-ggjt.bin stabilityai/stablelm-base-alpha-7b Q4_0 GGJT V3
stablelm-base-alpha-7b-q5_1.bin stabilityai/stablelm-base-alpha-7b Q5_1 GGML V3
stablelm-base-alpha-7b-q5_1-ggjt.bin stabilityai/stablelm-base-alpha-7b Q5_1 GGJT V3
stablelm-tuned-alpha-3b-f16.bin stabilityai/stablelm-tuned-alpha-3b F16 GGML V3
stablelm-tuned-alpha-3b-q4_0.bin stabilityai/stablelm-tuned-alpha-3b Q4_0 GGML V3
stablelm-tuned-alpha-3b-q4_0-ggjt.bin stabilityai/stablelm-tuned-alpha-3b Q4_0 GGJT V3
stablelm-tuned-alpha-3b-q5_1.bin stabilityai/stablelm-tuned-alpha-3b Q5_1 GGML V3
stablelm-tuned-alpha-3b-q5_1-ggjt.bin stabilityai/stablelm-tuned-alpha-3b Q5_1 GGJT V3
stablelm-tuned-alpha-7b-f16.bin stabilityai/stablelm-tuned-alpha-7b F16 GGML V3
stablelm-tuned-alpha-7b-q4_0.bin stabilityai/stablelm-tuned-alpha-7b Q4_0 GGML V3
stablelm-tuned-alpha-7b-q4_0-ggjt.bin stabilityai/stablelm-tuned-alpha-7b Q4_0 GGJT V3
stablelm-tuned-alpha-7b-q5_1.bin stabilityai/stablelm-tuned-alpha-7b Q5_1 GGML V3
stablelm-tuned-alpha-7b-q5_1-ggjt.bin stabilityai/stablelm-tuned-alpha-7b Q5_1 GGJT V3

Usage

Python via llm-rs:

Installation

Via pip: pip install llm-rs

Run inference

from llm_rs import AutoModel

#Load the model, define any model you like from the list above as the `model_file`
model = AutoModel.from_pretrained("rustformers/stablelm-ggml",model_file="stablelm-base-alpha-3b-q4_0-ggjt.bin")

#Generate
print(model.generate("The meaning of life is"))

Rust via Rustformers/llm:

Installation

git clone --recurse-submodules https://github.com/rustformers/llm.git
cd llm
cargo build --release

Run inference

cargo run --release -- gptneox infer -m path/to/model.bin  -p "Tell me how cool the Rust programming language is:"