Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf SanctumAI/OpenELM-3B-Instruct-GGUF:# Run inference directly in the terminal:
llama-cli -hf SanctumAI/OpenELM-3B-Instruct-GGUF:Use pre-built binary
# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf SanctumAI/OpenELM-3B-Instruct-GGUF:# Run inference directly in the terminal:
./llama-cli -hf SanctumAI/OpenELM-3B-Instruct-GGUF:Build from source code
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf SanctumAI/OpenELM-3B-Instruct-GGUF:# Run inference directly in the terminal:
./build/bin/llama-cli -hf SanctumAI/OpenELM-3B-Instruct-GGUF:Use Docker
docker model run hf.co/SanctumAI/OpenELM-3B-Instruct-GGUF:
This model was quantized by SanctumAI. To leave feedback, join our community in Discord.
OpenELM 3B Instruct GGUF
Model creator: apple
Original model: OpenELM-3B-Instruct
Model Summary:
Sachin Mehta, Mohammad Hossein Sekhavat, Qingqing Cao, Maxwell Horton, Yanzi Jin, Chenfan Sun, Iman Mirzadeh, Mahyar Najibi, Dmitry Belenko, Peter Zatloukal, Mohammad Rastegari
We introduce OpenELM, a family of Open Efficient Language Models. OpenELM uses a layer-wise scaling strategy to efficiently allocate parameters within each layer of the transformer model, leading to enhanced accuracy. We pretrained OpenELM models using the CoreNet library. We release both pretrained and instruction tuned models with 270M, 450M, 1.1B and 3B parameters. We release the complete framework, encompassing data preparation, training, fine-tuning, and evaluation procedures, alongside multiple pre-trained checkpoints and training logs, to facilitate open research.
Our pre-training dataset contains RefinedWeb, deduplicated PILE, a subset of RedPajama, and a subset of Dolma v1.6, totaling approximately 1.8 trillion tokens. Please check license agreements and terms of these datasets before using them.
Prompt Template:
If you're using Sanctum app, simply use Zephyr model preset.
Prompt template:
<|system|>
{system_prompt}</s>
<|user|>
{prompt}</s>
<|assistant|>
Hardware Requirements Estimate
| Name | Quant method | Size | Memory (RAM, vRAM) required |
|---|---|---|---|
| openelm-3b-instruct.Q2_K.gguf | Q2_K | 1.15 GB | 3.14 GB |
| openelm-3b-instruct.Q3_K_S.gguf | Q3_K_S | 1.34 GB | 3.32 GB |
| openelm-3b-instruct.Q3_K_M.gguf | Q3_K_M | 1.53 GB | 3.50 GB |
| openelm-3b-instruct.Q3_K_L.gguf | Q3_K_L | 1.66 GB | 3.61 GB |
| openelm-3b-instruct.Q4_0.gguf | Q4_0 | 1.74 GB | 3.68 GB |
| openelm-3b-instruct.Q4_K_S.gguf | Q4_K_S | 1.74 GB | 3.69 GB |
| openelm-3b-instruct.Q4_K_M.gguf | Q4_K_M | 1.89 GB | 3.83 GB |
| openelm-3b-instruct.Q4_K.gguf | Q4_K | 1.89 GB | 3.83 GB |
| openelm-3b-instruct.Q4_1.gguf | Q4_1 | 1.92 GB | 3.85 GB |
| openelm-3b-instruct.Q5_0.gguf | Q5_0 | 2.10 GB | 4.03 GB |
| openelm-3b-instruct.Q5_K_S.gguf | Q5_K_S | 2.10 GB | 4.03 GB |
| openelm-3b-instruct.Q5_K_M.gguf | Q5_K_M | 2.21 GB | 4.12 GB |
| openelm-3b-instruct.Q5_K.gguf | Q5_K | 2.21 GB | 4.12 GB |
| openelm-3b-instruct.Q5_1.gguf | Q5_1 | 2.29 GB | 4.20 GB |
| openelm-3b-instruct.Q6_K.gguf | Q6_K | 2.49 GB | 4.39 GB |
| openelm-3b-instruct.Q8_0.gguf | Q8_0 | 3.23 GB | 5.07 GB |
| openelm-3b-instruct.f16.gguf | f16 | 6.07 GB | 7.72 GB |
Disclaimer
Sanctum is not the creator, originator, or owner of any Model featured in the Models section of the Sanctum application. Each Model is created and provided by third parties. Sanctum does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Model listed there. You understand that supported Models can produce content that might be offensive, harmful, inaccurate or otherwise inappropriate, or deceptive. Each Model is the sole responsibility of the person or entity who originated such Model. Sanctum may not monitor or control the Models supported and cannot, and does not, take responsibility for any such Model. Sanctum disclaims all warranties or guarantees about the accuracy, reliability or benefits of the Models. Sanctum further disclaims any warranty that the Model will meet your requirements, be secure, uninterrupted or available at any time or location, or error-free, viruses-free, or that any errors will be corrected, or otherwise. You will be solely responsible for any damage resulting from your use of or access to the Models, your downloading of any Model, or use of any other Model provided by or through Sanctum.
- Downloads last month
- 8,793
2-bit
3-bit
4-bit
5-bit
6-bit
8-bit
16-bit
Install from brew
# Start a local OpenAI-compatible server with a web UI: llama-server -hf SanctumAI/OpenELM-3B-Instruct-GGUF:# Run inference directly in the terminal: llama-cli -hf SanctumAI/OpenELM-3B-Instruct-GGUF: