Instructions to use marianna13/prelude-9b-midtrain-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use marianna13/prelude-9b-midtrain-v1 with Transformers:

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

pipe = pipeline("text-generation", model="marianna13/prelude-9b-midtrain-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("marianna13/prelude-9b-midtrain-v1")
model = AutoModelForCausalLM.from_pretrained("marianna13/prelude-9b-midtrain-v1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use marianna13/prelude-9b-midtrain-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "marianna13/prelude-9b-midtrain-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "marianna13/prelude-9b-midtrain-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/marianna13/prelude-9b-midtrain-v1

SGLang

How to use marianna13/prelude-9b-midtrain-v1 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 "marianna13/prelude-9b-midtrain-v1" \
    --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": "marianna13/prelude-9b-midtrain-v1",
		"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 "marianna13/prelude-9b-midtrain-v1" \
        --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": "marianna13/prelude-9b-midtrain-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use marianna13/prelude-9b-midtrain-v1 with Docker Model Runner:
```
docker model run hf.co/marianna13/prelude-9b-midtrain-v1
```

Prelude 9B mid-train v1 — `stage4-128k-step1001`

Long-context mid-training of the Prelude 9B base (Qwen3 architecture, vocab 262,272), trained with Megatron on MareNostrum 5 and exported to Hugging Face format. Source checkpoint: prelude-9b-midtrain-v1-stage4-128k-yarn2-step1001.

This revision: 128K, YaRN x2 over 64K, RoPE theta 9M. Short 128K stage from stage 3 with YaRN x2 applied in training: 8,012 rows / 1.0B seen tokens, 1001 steps.

Each stage is a separate revision (branch) of this repo; main is the final stage (stage 4):

Revision	Context
`stage1-16k-step607`	16K (max_position 65536), RoPE theta 9M
`stage2-32k-step1821`	32K (max_position 65536), RoPE theta 9M
`stage3-64k-step4419`	64K (max_position 65536), RoPE theta 9M
`stage4-128k-step1001`	128K, YaRN x2 over 64K, RoPE theta 9M	this revision

from transformers import AutoModelForCausalLM, AutoTokenizer
m = AutoModelForCausalLM.from_pretrained("marianna13/prelude-9b-midtrain-v1", revision="stage4-128k-step1001", torch_dtype="bfloat16")

This is a base (not chat-tuned) model; it needs the BOS token. The chat template in the repo is the Prelude ChatML template used for later SFT.

Downloads last month: -

Safetensors

Model size

9B params

Tensor type

BF16

Model tree for marianna13/prelude-9b-midtrain-v1

Finetunes

1 model

Prelude 9B mid-train v1 — stage4-128k-step1001

Model tree for marianna13/prelude-9b-midtrain-v1

Prelude 9B mid-train v1 — `stage4-128k-step1001`