Instructions to use CladeTeam/CENO-600M-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use CladeTeam/CENO-600M-base with Transformers:

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

pipe = pipeline("text-generation", model="CladeTeam/CENO-600M-base", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("CladeTeam/CENO-600M-base", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use CladeTeam/CENO-600M-base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "CladeTeam/CENO-600M-base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "CladeTeam/CENO-600M-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/CladeTeam/CENO-600M-base

SGLang

How to use CladeTeam/CENO-600M-base 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 "CladeTeam/CENO-600M-base" \
    --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": "CladeTeam/CENO-600M-base",
		"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 "CladeTeam/CENO-600M-base" \
        --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": "CladeTeam/CENO-600M-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use CladeTeam/CENO-600M-base with Docker Model Runner:
```
docker model run hf.co/CladeTeam/CENO-600M-base
```

CENO-600M-base / README.md

Mishamq

Align CENO inference code and document numerical reference

2c457d7 verified 6 days ago

preview code

Raw History Blame Contribute Delete

4.2 kB

---
license: apache-2.0
library_name: transformers
tags:
  - dna
  - genomics
  - dna-language-model
  - mamba
  - moe
pipeline_tag: text-generation
---
# CENO-600M-base

**CENO-600M-base** is the base pretraining checkpoint of the 600M **CENO** DNA foundation model — a causal language model over genomic sequence built on a
Nemotron-H **Mamba / Attention / Mixture-of-Experts** hybrid backbone (no MSA inputs).

It is part of the **CENO** DNA foundation model family. Model code, the VEP pipeline, and a
generation demo live in the companion [CENO code repository](https://github.com/CladeTeam/CENO).
This checkpoint is standalone-loadable with `trust_remote_code=True` — the model code is
bundled here.

## Model details

|  |  |
|---|---|
| Family | CENO (base) |
| Training stage | Base pretraining (stage 2) |
| Parameters | 630.8M |
| Precision | bfloat16 |
| `model_type` | `ceno` |
| Architecture class | `CENOForCausalLM` |
| Auto-map (model) | `modeling_ceno.CENOForCausalLM` |
| Auto-map (tokenizer) | `ceno_tokenizer.CENOCharLevelTokenizer` |

## Architecture

| Property | Value |
|---|---|
| Hidden layers | 20 |
| Hidden size | 1024 |
| Attention heads | 16 |
| Intermediate size | 4096 |
| Experts (MoE) | 8 (top-2 per token) |
| Vocabulary | 512 (byte / character-level) |

The backbone is a Mamba / Attention / Mixture-of-Experts hybrid (Nemotron-H
architecture). The tokenizer is character-level, mapping DNA bases to their ASCII
byte codes.

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
ckpt = "CladeTeam/CENO-600M-base"
model = AutoModelForCausalLM.from_pretrained(ckpt, trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(ckpt, trust_remote_code=True)
ids = tokenizer.encode("ATCGATCG", return_tensors="pt")
# out = model.generate(ids, max_new_tokens=128)   # needs a CUDA GPU (Mamba kernels)
```

> The Mamba layers require CUDA kernels, so forward passes and generation need a GPU.
> Config, tokenizer, and weight loading are CPU-safe.

## Intended use

- **Base checkpoints (`CENO-*`)** — genomic-sequence generation and embedding extraction;
  downstream adaptation (fine-tuning, probing) for genomics tasks.
- **MSA checkpoints (`CENO-P-*`)** — variant effect prediction (VEP) by scoring wild-type
  vs. variant sequences with delta log-likelihood. See the TraitGym VEP example in the
  [CENO code repository](https://github.com/CladeTeam/CENO).

## License

Apache-2.0. The bundled model code is derived from NVIDIA's Nemotron-H Hugging Face
implementation (Apache-2.0); the tokenizer is derived from the Arc Institute Evo2
`CharLevelTokenizer` (Apache-2.0). See the `LICENSE` and `NOTICE` files in this repository
for full attribution.


<!-- CENO_INFERENCE_CURRENT -->
## Current inference implementation

Model weights and token IDs are unchanged. Current source and usage instructions
are maintained at [CladeTeam/CENO](https://github.com/CladeTeam/CENO).
Use the updated model revision instead of historical diagnostic archives.

CUDA BF16 inference uses the corrected Mamba, MoE, normalization and attention
arithmetic selected by this model's explicit reference configuration. Loading
uses the checkpoint dtype by default; explicit caller dtype overrides are honored.
Use `torch_dtype="auto"`, `.cuda().eval()` and `torch.inference_mode()` for raw
forward calls. `generate()` enters no-grad mode automatically.

The HF precision path is single-GPU, unpadded, and permits at most 131072 tokens
per prefill. Use the repository's vLLM entry for larger prefills. CPU, explicit
FP32 and explicit eager attention retain ordinary compatible behavior and are
not covered by the BF16 bitwise-reference claim. Different Megatron CP references
are not interchangeable. The long 1M reference evidence belongs to CENO-1B-1m;
short multi-size checks do not imply the same long coverage for every model.

The tested HF stack is PyTorch 2.7.0a0 (NVIDIA 25.03), Transformers 4.48.3,
mamba-ssm 2.2.4, causal-conv1d 1.2.2.post1 and FlashAttention 2.7.3. HF and vLLM
use separate environments. Numerical correctness is not a guarantee of
repetition-free or biologically functional generation.
<!-- /CENO_INFERENCE_CURRENT -->

	---
	license: apache-2.0
	library_name: transformers
	tags:
	- dna
	- genomics
	- dna-language-model
	- mamba
	- moe
	pipeline_tag: text-generation
	---

	# CENO-600M-base

	CENO-600M-base is the base pretraining checkpoint of the 600M CENO DNA foundation model — a causal language model over genomic sequence built on a
	Nemotron-H Mamba / Attention / Mixture-of-Experts hybrid backbone (no MSA inputs).

	It is part of the CENO DNA foundation model family. Model code, the VEP pipeline, and a
	generation demo live in the companion [CENO code repository](https://github.com/CladeTeam/CENO).
	This checkpoint is standalone-loadable with `trust_remote_code=True` — the model code is
	bundled here.

	## Model details

	\| \| \|
	\|---\|---\|
	\| Family \| CENO (base) \|
	\| Training stage \| Base pretraining (stage 2) \|
	\| Parameters \| 630.8M \|
	\| Precision \| bfloat16 \|
	\| `model_type` \| `ceno` \|
	\| Architecture class \| `CENOForCausalLM` \|
	\| Auto-map (model) \| `modeling_ceno.CENOForCausalLM` \|
	\| Auto-map (tokenizer) \| `ceno_tokenizer.CENOCharLevelTokenizer` \|

	## Architecture

	\| Property \| Value \|
	\|---\|---\|
	\| Hidden layers \| 20 \|
	\| Hidden size \| 1024 \|
	\| Attention heads \| 16 \|
	\| Intermediate size \| 4096 \|
	\| Experts (MoE) \| 8 (top-2 per token) \|
	\| Vocabulary \| 512 (byte / character-level) \|

	The backbone is a Mamba / Attention / Mixture-of-Experts hybrid (Nemotron-H
	architecture). The tokenizer is character-level, mapping DNA bases to their ASCII
	byte codes.

	## Usage

	```python
	from transformers import AutoModelForCausalLM, AutoTokenizer

	ckpt = "CladeTeam/CENO-600M-base"
	model = AutoModelForCausalLM.from_pretrained(ckpt, trust_remote_code=True)
	tokenizer = AutoTokenizer.from_pretrained(ckpt, trust_remote_code=True)

	ids = tokenizer.encode("ATCGATCG", return_tensors="pt")
	# out = model.generate(ids, max_new_tokens=128) # needs a CUDA GPU (Mamba kernels)
	```

	> The Mamba layers require CUDA kernels, so forward passes and generation need a GPU.
	> Config, tokenizer, and weight loading are CPU-safe.

	## Intended use

	- *Base checkpoints (`CENO-`)** — genomic-sequence generation and embedding extraction;
	downstream adaptation (fine-tuning, probing) for genomics tasks.
	- *MSA checkpoints (`CENO-P-`)** — variant effect prediction (VEP) by scoring wild-type
	vs. variant sequences with delta log-likelihood. See the TraitGym VEP example in the
	[CENO code repository](https://github.com/CladeTeam/CENO).

	## License

	Apache-2.0. The bundled model code is derived from NVIDIA's Nemotron-H Hugging Face
	implementation (Apache-2.0); the tokenizer is derived from the Arc Institute Evo2
	`CharLevelTokenizer` (Apache-2.0). See the `LICENSE` and `NOTICE` files in this repository
	for full attribution.


	<!-- CENO_INFERENCE_CURRENT -->
	## Current inference implementation

	Model weights and token IDs are unchanged. Current source and usage instructions
	are maintained at [CladeTeam/CENO](https://github.com/CladeTeam/CENO).
	Use the updated model revision instead of historical diagnostic archives.

	CUDA BF16 inference uses the corrected Mamba, MoE, normalization and attention
	arithmetic selected by this model's explicit reference configuration. Loading
	uses the checkpoint dtype by default; explicit caller dtype overrides are honored.
	Use `torch_dtype="auto"`, `.cuda().eval()` and `torch.inference_mode()` for raw
	forward calls. `generate()` enters no-grad mode automatically.

	The HF precision path is single-GPU, unpadded, and permits at most 131072 tokens
	per prefill. Use the repository's vLLM entry for larger prefills. CPU, explicit
	FP32 and explicit eager attention retain ordinary compatible behavior and are
	not covered by the BF16 bitwise-reference claim. Different Megatron CP references
	are not interchangeable. The long 1M reference evidence belongs to CENO-1B-1m;
	short multi-size checks do not imply the same long coverage for every model.

	The tested HF stack is PyTorch 2.7.0a0 (NVIDIA 25.03), Transformers 4.48.3,
	mamba-ssm 2.2.4, causal-conv1d 1.2.2.post1 and FlashAttention 2.7.3. HF and vLLM
	use separate environments. Numerical correctness is not a guarantee of
	repetition-free or biologically functional generation.
	<!-- /CENO_INFERENCE_CURRENT -->