Instructions to use TroyDoesAI/Codestral-21B-Pruned with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TroyDoesAI/Codestral-21B-Pruned with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="TroyDoesAI/Codestral-21B-Pruned")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("TroyDoesAI/Codestral-21B-Pruned")
model = AutoModelForCausalLM.from_pretrained("TroyDoesAI/Codestral-21B-Pruned")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use TroyDoesAI/Codestral-21B-Pruned with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TroyDoesAI/Codestral-21B-Pruned"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TroyDoesAI/Codestral-21B-Pruned",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TroyDoesAI/Codestral-21B-Pruned

SGLang

How to use TroyDoesAI/Codestral-21B-Pruned with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "TroyDoesAI/Codestral-21B-Pruned" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TroyDoesAI/Codestral-21B-Pruned",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "TroyDoesAI/Codestral-21B-Pruned" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TroyDoesAI/Codestral-21B-Pruned",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TroyDoesAI/Codestral-21B-Pruned with Docker Model Runner:
```
docker model run hf.co/TroyDoesAI/Codestral-21B-Pruned
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

For those trying to shoe horn this large model on your machine every GB of saved memory counts when offloading to System RAM!

Here is a pruned down the 22.2 Billion parameter model by 2 junk layers to make a 21.5B that doesnt appear to lose any sense of quality.

https://huggingface.co/mistralai/Codestral-22B-v0.1

Downloads last month: 8

Safetensors

Model size

21B params

Tensor type

F16

Model tree for TroyDoesAI/Codestral-21B-Pruned

Quantizations

2 models