use hf instead of huggingface-cli; add debug info

Author: cquil11

benchmarks/dsr1_fp8_h200_slurm.sh

@@ -17,7 +17,7 @@
 echo "JOB \$SLURM_JOB_ID running on \$SLURMD_NODENAME"
 
 pip3 install --user sentencepiece
-huggingface-cli download $MODEL
+hf download $MODEL
 PORT=$(( 8888 + $PORT_OFFSET ))
 SERVER_LOG=$(mktemp /tmp/server-XXXXXX.log)
 

benchmarks/dsr1_fp8_mi300x_slurm.sh

@@ -15,7 +15,7 @@
 
 echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
 
-huggingface-cli download $MODEL
+hf download $MODEL
 
 SERVER_LOG=$(mktemp /tmp/server-XXXXXX.log)
 PORT=8888

benchmarks/dsr1_fp8_mi325x_slurm.sh

@@ -2,9 +2,15 @@
 
 echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
 
+echo "=== CPU Information Inside Container ==="
+nproc
+cat /proc/cpuinfo | grep processor | wc -l
+cat /sys/fs/cgroup/cpuset/cpuset.cpus 2>/dev/null || echo "cpuset info not available"
+echo "========================================"
+
 SERVER_LOG=$(mktemp /tmp/server-XXXXXX.log)
 PORT=8888
-huggingface-cli download $MODEL
+hf download $MODEL
 
 # Reference
 # https://rocm.docs.amd.com/en/docs-7.0-rc1/preview/benchmark-docker/inference-sglang-deepseek-r1-fp8.html#run-the-inference-benchmark

benchmarks/gptoss_fp4_mi300x_slurm.sh

@@ -15,7 +15,7 @@
 
 echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
 
-huggingface-cli download $MODEL
+hf download $MODEL
 
 SERVER_LOG=$(mktemp /tmp/server-XXXXXX.log)
 PORT=8888

benchmarks/gptoss_fp4_mi325x_slurm.sh

@@ -16,7 +16,7 @@
 
 echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
 
-huggingface-cli download $MODEL
+hf download $MODEL
 
 SERVER_LOG=$(mktemp /tmp/server-XXXXXX.log)
 PORT=8888