Views
No views yet
bc1qsvfduzj9fjs9fugpc52yver3f2g8fp7xjxecdvenable_thinking
reasoning mode — so this checkpoint is a drop-in replacement for the original at the architecture level
and serves out-of-the-box on vLLM.</think>), then baked into the weights as an offline orthogonal projection on the residual-write
modules — using our own custom abliteration framework that operates directly on the packed NVFP4
tensors (dequantize → project → requantize), with no full-precision decompress and no training.enable_thinking works)| Property | Value |
|---|---|
| Architecture | NemotronHForCausalLM (model_type: nemotron_h) |
| Total / Active Parameters | 550B / 55B |
| Layers | 108 — 48 Mamba-2 · 48 Latent-MoE · 12 Attention (hybrid) |
| Hidden Size | 8192 |
| Routed / Shared Experts | 512 routed (22 active/token, 2048-dim latent space) · 1 shared |
| Attention | 64 heads / 2 KV heads |
| Multi-Token Prediction | 1 MTP layer (native speculative decoding) |
| Vocabulary | 131,072 |
| Context Length | up to 1M tokens (256K default) |
| Quantization | NVFP4 (modelopt, mixed-precision: select layers FP8/BF16) |
config.json, model.safetensors.index.json, tokenizer,
chat_template.jinja, generation_config.json. Total on disk: ~329 GB.1vllm serve OpenYourMind/OpenYourMind-NVIDIA-Nemotron-3-Ultra-550B-A55B-abliterated-uncensored-NVFP4 \
2 --trust-remote-code \
3 --tensor-parallel-size 4 \
4 --enable-expert-parallel \
5 --max-model-len 262144 \
6 --reasoning-parser nemotron_v3--tensor-parallel-size 2. Requires vLLM ≥ 0.22.1from openai import OpenAI
2client = OpenAI(base_url="http://localhost:8000/v1", api_key="x")
3r = client.chat.completions.create(
4 model="OpenYourMind/OpenYourMind-NVIDIA-Nemotron-3-Ultra-550B-A55B-abliterated-uncensored-NVFP4",
5 messages=[{"role": "user", "content": "Your prompt here"}],
6 extra_body={"chat_template_kwargs": {"enable_thinking": True}},
7)
8print(r.choices[0].message.content)temperature=1.0, top_p=0.95 (the values in generation_config.json). A mild
repetition_penalty (~1.1) is recommended for long generations.enable_thinking=True in chat_template_kwargs; reasoning streams inside
<think>…</think> before the answer. Do not feed previous-turn reasoning back into multi-turn history.