sarman/dgx-spark-playbooks
1
0
Fork 0
mirror of https://github.com/NVIDIA/dgx-spark-playbooks.git synced 2026-04-26 20:03:52 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge ce1609e892 into 3ba4d58f1e

Browse Source
This commit is contained in:
Shakhizat Nurgaliyev 2026-04-20 11:44:54 +00:00 committed by GitHub
commit c2519e9cb0
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
nvidia/nemotron/README.md
View File

@ -101,7 +101,7 @@ Build llama.cpp with CUDA enabled and targeting the GB10's sm_121 compute archit
```bash ```bash
mkdir build && cd build mkdir build && cd build
cmake .. -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES="121" -DLLAMA_CURL=OFF cmake .. -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES="121" -DLLAMA_CURL=OFF -DGGML_CUDA_FA_ALL_QUANTS=ON
make -j8 make -j8
``` ```
@ -128,6 +128,7 @@ Launch the inference server with the Nemotron model. The server provides an Open
--model ~/models/nemotron3-gguf/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL.gguf \ --model ~/models/nemotron3-gguf/Nemotron-3-Nano-30B-A3B-UD-Q8_K_XL.gguf \
--host 0.0.0.0 \ --host 0.0.0.0 \
--port 30000 \ --port 30000 \
--flash-attn 1 \
--n-gpu-layers 99 \ --n-gpu-layers 99 \
--ctx-size 8192 \ --ctx-size 8192 \
--threads 8 --threads 8
@ -136,6 +137,7 @@ Launch the inference server with the Nemotron model. The server provides an Open
**Parameter explanation:** **Parameter explanation:**
- `--host 0.0.0.0`: Listen on all network interfaces - `--host 0.0.0.0`: Listen on all network interfaces
- `--port 30000`: API server port - `--port 30000`: API server port
- `--flash-attn 1`: Enables Flash Attention
- `--n-gpu-layers 99`: Offload all layers to GPU - `--n-gpu-layers 99`: Offload all layers to GPU
- `--ctx-size 8192`: Context window size (can increase up to 1M) - `--ctx-size 8192`: Context window size (can increase up to 1M)
- `--threads 8`: CPU threads for non-GPU operations - `--threads 8`: CPU threads for non-GPU operations