Instructions to use crumb/d1536-250MT-full with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use crumb/d1536-250MT-full with Transformers:

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

pipe = pipeline("text-generation", model="crumb/d1536-250MT-full", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("crumb/d1536-250MT-full", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use crumb/d1536-250MT-full with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "crumb/d1536-250MT-full"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "crumb/d1536-250MT-full",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/crumb/d1536-250MT-full

SGLang

How to use crumb/d1536-250MT-full 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 "crumb/d1536-250MT-full" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "crumb/d1536-250MT-full",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "crumb/d1536-250MT-full" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "crumb/d1536-250MT-full",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use crumb/d1536-250MT-full with Docker Model Runner:
```
docker model run hf.co/crumb/d1536-250MT-full
```

model	avg	arc	truthfulqa	hellaswag	mmlu
crumb/d1536-250MT-full	30.61	23.29	47.1	26.60	25.45787898435144

trained for 250MT/16GT, compared to GPT-2 suite: between gpt2-124m and gpt2-345m (the params are between as well). 254,883,841 non-embedding parameters, this is a roughly 1Tok/Param initial saturation of weights to be further pretrained with ReLora.

Downloads last month: 6