No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from MultiverseComputingCAI/Pulsar-16B-BF16
All models by MultiverseComputingCAI →
[blocked]
Powered by CompactifAI
Optimized for Fast and Efficient Inference · Reduced Memory Footprint
Pulsar 16B is a model based on NVIDIA-Nemotron-3-Nano-30B-A3B-BF16, developed by Multiverse Computing. The original model is a ~31.6B parameter, part of the Nemotron model family. It supports long-context inference up to 1M tokens and is designed for general-purpose language modeling tasks.
This version applies model compression techniques to significantly reduce parameter count and deployment requirements while maintaining compatibility with the Nemotron Hybrid Mamba2-Transformer with MoE architecture. The resulting model achieves 50% compression, reducing the parameter count to 16.15B parameters and lowering memory requirements.
This model can be loaded with the Transformers API. Use trust_remote_code=True. Recommended approach: AutoModelForCausalLM with apply_chat_template. This configuration has been tested with Transformers 4.57.6.
Alternatively you can use the pipeline API with trust_remote_code=True; the pipeline returns the full conversation structure, so extract the assistant message from outputs[0]["generated_text"] as needed.
Pulsar 16B uses the same Nemotron v3 reasoning tags as the base model. NVIDIA provides the vLLM plugin as nano_v3_reasoning_parser.py on the base Hugging Face repo (not specific to Pulsar). Direct download:
You can keep any local filename; the vllm serve flags below assume the file is in the current directory as nano_v3_reasoning_parser.py. If you mirror an identical copy under the Pulsar model repo, use that URL instead.
Note: The NeMo container
nvcr.io/nvidia/nemo:25.11.nemotron_3_nanocomes withmamba_ssmandcausal-conv1dpre-installed.
Pulsar 16B supports a hybrid reasoning mode: the model can either think step-by-step before answering (reasoning mode) or reply directly (non-reasoning mode). The behaviour is controlled via the enable_thinking flag in the chat template.
This section provides a brief overview of reasoning control in Pulsar 16B. For comprehensive details please see the official Nemotron-3 Nano-30B model card at: https://build.nvidia.com/nvidia/nemotron-3-nano-30b-a3b/modelcard
Pass enable_thinking through apply_chat_template:
Thinking ON (default)
Thinking OFF
When thinking is ON the model opens a <think> block before the answer.
Set the default for all requests at startup with --default-chat-template-kwargs.
Requires recent versions of vLLM.
Thinking OFF for all requests
Thinking ON for all requests (default if flag is omitted)
--trust-request-chat-templateis required to allow per-request overrides.
Individual requests can override the server default by passing chat_template_kwargs in the request body. This works regardless of the server-level default.
Thinking ON/OFF for one request
Pulsar 16B emits tool calls in the following format:
When serving (e.g with vLLM), you must use the qwen3_coder tool parser.
The base model nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 is a large language model (LLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks. The model's reasoning capabilities can be configured through a flag in the chat template. See the original model card for details.
CompactifAI was applied to produce a smaller, efficient model (16B parameters) while aiming to preserve reasoning and tool-use capabilities. Supervised Fine Tuning was applied for improving cabapilities.
[Image blocked: Combined benchmark chart]
[Image blocked: Quantization results]
[Image blocked: Performance results]
temperature: 0.0, top_p: 1.0Pulsar 16B preserves strong long-context behavior after compression, tracking the Nemotron-3-Nano-30B-A3B baseline closely across retrieval-heavy and full-suite long-context evaluations. Results are reported for LongBench v1, AA-LCR, NIAH, and RULER groupings up to 256k context.
[Image blocked: Long-context benchmark results]
Benchmark scores were obtained with the following setups. Methodology varies by benchmark family.
temp 1.0, top_p 1.0temp: 1.0, top_p: 1.0, reasoning_effort: hightemp: 0.6, top_p: 0.95, top_k: 20, min_p: 0.0temp: 0.15Trained mainly on English with added Spanish. No systematic evaluation for languages outside English and Spanish.
If you use this model, please cite the base model and Pulsar 16B:
Built by Multiverse Computing · Report an issue [blocked] · Discord
import torchfrom transformers import AutoTokenizer, AutoModelForCausalLM
model_id = "MultiverseComputingCAI/Pulsar-16B-BF16"
tokenizer = AutoTokenizer.from_pretrained( model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained( model_id, device_map="cuda" if torch.cuda.is_available() else "auto", torch_dtype=torch.bfloat16, trust_remote_code=True,)messages = [ {"role": "user", "content": "Write a haiku about GPUs"},]
tokenized_chat = tokenizer.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_tensors="pt").to(model.device)
outputs = model.generate( tokenized_chat, max_new_tokens=1024, temperature=1.0, top_p=1.0, eos_token_id=tokenizer.eos_token_id)print(tokenizer.decode(outputs[0]))pip install -U "vllm>=0.12.0"wget https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16/resolve/main/nano_v3_reasoning_parser.pyvllm serve MultiverseComputingCAI/Pulsar-16B-BF16 \ --served-model-name model \ --max-num-seqs 8 \ --tensor-parallel-size 1 \ --port 8000 \ --trust-remote-code \ --enable-auto-tool-choice \ --tool-call-parser qwen3_coder \ --reasoning-parser-plugin nano_v3_reasoning_parser.py \ --reasoning-parser nano_v3tokenized_chat = tokenizer.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_tensors="pt", enable_thinking=True, # default — can be omitted)tokenized_chat = tokenizer.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_tensors="pt", enable_thinking=False,)output = tokenizer.decode(outputs[0], skip_special_tokens=True)# Split on </think> to separate reasoning from the final answerif "</think>" in output: reasoning, answer = output.split("</think>", 1) reasoning = reasoning.replace("<think>", "").strip() answer = answer.strip()else: answer = outputvllm serve MultiverseComputingCAI/Pulsar-16B-BF16 \ --served-model-name model \ --reasoning-parser-plugin nano_v3_reasoning_parser.py \ --reasoning-parser nano_v3 \ --trust-request-chat-template \ --default-chat-template-kwargs '{"enable_thinking": false}' \ ...vllm serve MultiverseComputingCAI/Pulsar-16B-BF16 \ --served-model-name model \ --reasoning-parser-plugin nano_v3_reasoning_parser.py \ --reasoning-parser nano_v3 \ --trust-request-chat-template \ --default-chat-template-kwargs '{"enable_thinking": true}' \ ...import requests
response = requests.post("http://localhost:8000/v1/chat/completions", json={ "model": "model", "messages": [{"role": "user", "content": "Solve: x² - 5x + 6 = 0"}], "max_tokens": 1024, "temperature": 1.0, "chat_template_kwargs": {"enable_thinking": True},})<tool_call><function=get_weather><parameter=city>Paris</parameter><parameter=unit>celsius</parameter></function></tool_call>vllm serve <model_path> \ --enable-auto-tool-choice \ --tool-call-parser qwen3_coder \ --trust-remote-code@misc{nemotron3nanoTR, title = {NVIDIA Nemotron 3 Nano Technical Report}, author = {{NVIDIA}}, year = {2025}, url = {https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Nano-Technical-Report.pdf}}@misc{nemotron3nanoslim16b, title = {Pulsar 16B: Model developed from NVIDIA Nemotron-3-Nano-30B-A3B}, author = {Multiverse Computing}, year = {2026}, url = {https://huggingface.co/MultiverseComputingCAI/Pulsar-16B-BF16}, note = {Model developed based on nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 using CompactifAI technology}}