feat(ux): better readme

apps/prepare-benchmark/.env.template

@@ -1,3 +1,5 @@
-# use absolute path
-DATA_DIR="../../data" # relative to this file 
+# For downloading datasets from Hugging Face
 HF_TOKEN="<your-huggingface-token>"
+
+# [Optional] Data loading directory, by default `../../data`
+DATA_DIR="../../data" # relative to this file 

apps/run-agent/.env.template

@@ -1,27 +1,31 @@
-# third party API keys
-SERPER_API_KEY=""
-JINA_API_KEY=""
-E2B_API_KEY=""
-
-# Google
-GEMINI_API_KEY=""
+# Using OpenRouter to provide primary agent model
+OPENROUTER_API_KEY=""
+OPENROUTER_BASE_URL="https://openrouter.ai/api/v1"
 
-# Anthropic
+# Anthropic, for vision tools
 ANTHROPIC_API_KEY=""
 ANTHROPIC_BASE_URL="https://api.anthropic.com"
 
-# openAI
+# OpenAI, for audio tools, intent recognition, and answer extraction
 OPENAI_API_KEY=""
 OPENAI_BASE_URL="https://api.openai.com/v1"
 
-# openrouter
-OPENROUTER_API_KEY=""
-OPENROUTER_BASE_URL="https://openrouter.ai/api/v1"
+# Gemini, for YouTube tasks
+GEMINI_API_KEY=""
+
+# Third party API keys
+# For Google search and website scraping
+SERPER_API_KEY=""
+# For website scraping
+JINA_API_KEY=""
+# For the Linux sandbox
+E2B_API_KEY=""
 
-# NewAPI
+# [Optional] NewAPI, alternative to OpenRouter 
 NEWAPI_API_KEY=""
 NEWAPI_BASE_URL=""
 
-# use HTTPS proxy
+# [Optional] for network proxy, null by default
 HTTPS_PROXY=""
+# [Optional] Data loading directory, by default `../../data`
 DATA_DIR="../../data"

apps/run-agent/scripts/claude-sonnet-3.7/run_evaluate_multiple_runs_gaia-validation.sh

@@ -9,6 +9,7 @@ BENCHMARK_NAME="gaia-validation"
 LLM_PROVIDER="claude_openrouter"
 LLM_MODEL="anthropic/claude-3.7-sonnet"
 AGENT_SET="miroflow"
+MAX_CONCURRENT=5
 
 RESULTS_DIR="logs/${BENCHMARK_NAME}/${LLM_PROVIDER}_${LLM_MODEL}_${AGENT_SET}"
 
@@ -32,7 +33,7 @@ for i in $(seq 1 $NUM_RUNS); do
             llm.model_name=$LLM_MODEL \
             llm.async_client=true \
             benchmark.execution.max_tasks=null \
-            benchmark.execution.max_concurrent=5 \
+            benchmark.execution.max_concurrent=$MAX_CONCURRENT \
             benchmark.execution.pass_at_k=1 \
             agent=$AGENT_SET \
             output_dir="$RESULTS_DIR/$RUN_ID" \