naver-hyperclovax
/

HyperCLOVAX-SEED-Vision-Instruct-3B

Text Generation

hyperclovax_vlm

Model card Files Files and versions

DongHyunKim commited on Sep 16

Commit

f0e805a

·

verified ·

1 Parent(s): d61f4cc

Update README.md

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -322,7 +322,7 @@ export ATTENTION_BACKEND=FLASH_ATTN_VLLM_V1
 VLLM_USE_V1=1 VLLM_ATTENTION_BACKEND=${ATTENTION_BACKEND} CUDA_VISIBLE_DEVICES=0,1 python -m vllm.entrypoints.openai.api_server \
     --seed 20250525 \
     --port ${NSML_PORT2} \
-    --allowed-local-media-path "/mnt/ocr-nfsx1/public/hodong.lee/cloned/vLLM/v0.8.4/vllm/ipynbs" \
     --max-model-len 8192 \
     --max-num-batched-tokens 8192 \
     --max-num-seqs 128 \
@@ -336,7 +336,7 @@ VLLM_USE_V1=1 VLLM_ATTENTION_BACKEND=${ATTENTION_BACKEND} CUDA_VISIBLE_DEVICES=0
     --dtype float16 \
     --trust-remote-code \
     --chat-template-content-format "openai" \
-    --download-dir "/mnt/ocr-nfsx1/public_datasets/.cache/huggingface/hub"
 ```
 **Request Example**:

 VLLM_USE_V1=1 VLLM_ATTENTION_BACKEND=${ATTENTION_BACKEND} CUDA_VISIBLE_DEVICES=0,1 python -m vllm.entrypoints.openai.api_server \
     --seed 20250525 \
     --port ${NSML_PORT2} \
+    --allowed-local-media-path $ALLOWED_LOCAL_MEDIA_PATH \
     --max-model-len 8192 \
     --max-num-batched-tokens 8192 \
     --max-num-seqs 128 \
     --dtype float16 \
     --trust-remote-code \
     --chat-template-content-format "openai" \
+    --download-dir $DONWLOAD_DIR
 ```
 **Request Example**: