[feat] run alfworld rollout script example

Author: realtmxi

scripts/rollout/run_alfworld.sh

@@ -0,0 +1,80 @@
+#!/bin/bash
+
+# Script to run AlfWorld rollout evaluation
+# This script runs the evaluation directly without Docker
+
+TIMESTAMP=$(date +"%Y%m%d_%H%M%S")
+LOG_DIR="logs_${TIMESTAMP}"
+RESULTS_DIR="results_${TIMESTAMP}"
+
+echo "Starting AlfWorld rollout evaluation"
+echo "Timestamp: ${TIMESTAMP}"
+echo "Results will be saved to: ${RESULTS_DIR}"
+
+# Create directories
+mkdir -p "${LOG_DIR}"
+mkdir -p "${RESULTS_DIR}"
+
+# Activate virtual environment if it exists
+if [ -f "/opt/openmanus-venv/bin/activate" ]; then
+    source /opt/openmanus-venv/bin/activate
+fi
+
+# Function to run a model
+run_model() {
+    local model_name=$1
+    local base_url=$2
+    local display_name=$3
+    local safe_name=$(echo "$display_name" | tr ' /' '_')
+    
+    echo "Starting ${display_name}..."
+    
+    # Build command
+    local cmd="python scripts/rollout/openmanus_rollout.py \
+        --env alfworld \
+        --unique_envs \
+        --batch_size 10 \
+        --concurrency 10 \
+        --total_envs 200 \
+        --history_length 30 \
+        --model '${model_name}' \
+        --chat_root '${RESULTS_DIR}/${safe_name}' \
+        --dump_path '${RESULTS_DIR}/${safe_name}/trajectory.jsonl'"
+    
+    if [ ! -z "$base_url" ]; then
+        cmd="${cmd} --base_url '${base_url}/v1'"
+    fi
+    
+    # Run with logging
+    echo "Command: ${cmd}" | tee -a "${LOG_DIR}/main.log"
+    eval "${cmd}" 2>&1 | tee "${LOG_DIR}/${safe_name}.log"
+    
+    echo "Completed ${display_name}" | tee -a "${LOG_DIR}/main.log"
+}
+
+# Main execution
+echo "Launching models..." | tee -a "${LOG_DIR}/main.log"
+
+# OpenAI models (uncomment to use)
+# run_model "gpt-4o" "" "GPT-4o"
+# run_model "gpt-4o-mini" "" "GPT-4o-mini"
+
+# vLLM models - run sequentially
+run_model "qwen3-8b" "http://129.212.187.116:8001" "Qwen3-8B"
+run_model "llama-3.1-8b-instruct" "http://129.212.176.75:8001" "Llama3.1-8B"
+run_model "qwen2.5-7b-instruct" "http://134.199.196.219:8001" "Qwen2.5-7B"
+run_model "qwen2.5-72b-instruct" "http://134.199.196.239:8001" "Qwen2.5-72B"
+run_model "llama-3.3-70b-instruct" "http://129.212.178.4:8001" "Llama3.3-70B"
+
+echo ""
+echo "========================================="
+echo "All models completed!"
+echo "========================================="
+echo ""
+echo "Logs are in: ${LOG_DIR}/"
+echo "Results are in: ${RESULTS_DIR}/"
+echo ""
+echo "To check results:"
+echo "  ls -la ${RESULTS_DIR}/"
+echo "To view logs:"
+echo "  tail -f ${LOG_DIR}/*.log"