// Run ClawGUI-Eval benchmarks on GUI grounding models: check environment, run inference with progress monitoring, judge predictions, compute metrics, and report results.
| name | clawgui-eval |
| description | Run ClawGUI-Eval benchmarks on GUI grounding models: check environment, run inference with progress monitoring, judge predictions, compute metrics, and report results. |
| metadata | {"nanobot":{"emoji":"📊","requires":{"bins":["python3","nvidia-smi"]}}} |
Evaluate GUI grounding models on standard benchmarks (ScreenSpot-Pro, ScreenSpot-v2, UIVision, MMBench-GUI, OSWorld-G, AndroidControl). The eval framework lives at clawgui-eval/ inside the workspace.
All commands below MUST be run with cd clawgui-eval first (relative paths for data/, image/, output/ depend on this).
Before running any evaluation, execute these checks and report results to the user.
nvidia-smi --query-gpu=index,name,memory.total,memory.free --format=csv,noheader
Count the lines to determine available GPU count. Use this as NUM_GPUS unless the user specifies otherwise.
free -h | head -3
df -h .
cd clawgui-eval && python3 -c "import torch; print(f'torch={torch.__version__}, cuda={torch.cuda.is_available()}, gpus={torch.cuda.device_count()}')"
python3 -c "import flash_attn; print(f'flash_attn={flash_attn.__version__}')" 2>&1
If flash_attn is NOT installed, you MUST clearly warn the user:
"FlashAttention-2 未安装。框架会自动降级为 SDPA（PyTorch 内置），可以正常运行但精度可能略有下降。强烈建议安装 FlashAttention-2 以获得最佳精度："
pip install flash-attn --no-build-isolation如果从源码编译太慢，可以从 https://github.com/Dao-AILab/flash-attention/releases 下载预编译 wheel 安装。
Always include this warning prominently in the environment check report if flash_attn is missing. Do not silently skip it.
ls clawgui-eval/data/ | head -5
ls clawgui-eval/image/ | head -3
Both data/ and image/ must exist. If image/ is missing, tell the user to download and unzip the dataset first.
If the user provides a local MODEL_PATH, verify it exists:
ls <MODEL_PATH>/config.json 2>/dev/null && echo "OK" || echo "NOT FOUND"
Report all results to the user. If any critical check fails (no GPU, no CUDA, missing data/image), stop and explain what is needed.
The full pipeline is: Infer → Judge → Metric. You MUST execute all three steps in sequence -- do NOT stop after Judge. The Metric step produces the final accuracy numbers that the user needs.
The user must provide:
MODEL_TYPE -- which model architecture (see script table below)MODEL_PATH -- HuggingFace ID or local directoryEXPERIMENT_NAME -- a label for this run (output goes to output/<EXPERIMENT_NAME>/...)BENCHMARK -- which benchmark to evaluate on (see naming rules below)Only the following MODEL_TYPE values are supported. If the user requests a model not in this list, do NOT proceed -- tell them the model is not yet supported by ClawGUI-Eval and show this table for reference.
| MODEL_TYPE | Example HuggingFace IDs (tested sizes) | SS-Pro | SS-V2 | UIVision | MMBench-GUI | OSWorld-G | AndroidControl |
|---|---|---|---|---|---|---|---|
qwen3vl | Qwen/Qwen3-VL-2B-Instruct, Qwen/Qwen3-VL-4B-Instruct, Qwen/Qwen3-VL-8B-Instruct | Y | Y | Y | Y | Y | Y |
qwen25vl | Qwen/Qwen2.5-VL-3B-Instruct, Qwen/Qwen2.5-VL-7B-Instruct | Y | Y | Y | Y | Y | Y |
maiui | Tongyi-MAI/MAI-UI-2B, Tongyi-MAI/MAI-UI-8B | Y | Y | Y | Y | Y | - |
uitars | ByteDance-Seed/UI-TARS-1.5-7B | Y | Y | Y | Y | Y | - |
uivenus15 | inclusionAI/UI-Venus-1.5-2B, inclusionAI/UI-Venus-1.5-8B | Y | Y | Y | Y | Y | - |
guiowl15 | mPLUG/GUI-Owl-1.5-2B-Instruct, mPLUG/GUI-Owl-1.5-4B-Instruct, mPLUG/GUI-Owl-1.5-8B-Instruct | Y | Y | Y | Y | Y | - |
guig2 | inclusionAI/GUI-G2-7B | Y | Y | Y | Y | Y | - |
stepgui | stepfun-ai/GELab-Zero-4B-preview | Y | Y | Y | Y | Y | - |
uivenus | inclusionAI/UI-Venus-Ground-7B | Y | Y | Y | Y | Y | - |
- means the combination is not supported. If the user requests a MODEL_TYPE + BENCHMARK combination marked -, refuse and explain it is not available.
Users can substitute different sizes within the same model family by changing the size in the HuggingFace ID (e.g. Qwen3-VL-2B → Qwen3-VL-4B). The MODEL_TYPE stays the same regardless of model size. Users can also provide a local path instead of a HuggingFace ID.
API-only models (gemini, seed) are NOT supported by this skill -- they require the API/vLLM backend which is outside scope.
Script templates are at clawgui-eval/scripts/infer/transformers/. Pick the one matching the MODEL_TYPE:
| Template script | MODEL_TYPE |
|---|---|
qwen3vl_run_transformers.sh | qwen3vl |
qwen25vl_run_transformers.sh | qwen25vl |
maiui_run_transformers.sh | maiui |
uitars_run_transformers.sh | uitars |
guiowl15_run_transformers.sh | guiowl15 |
guig2_run_transformers.sh | guig2 |
stepgui_run_transformers.sh | stepgui |
uivenus15_run_transformers.sh | uivenus15 |
uivenus_run_transformers.sh | uivenus |
qwen3vl_run_transformers_androidcontrol.sh | qwen3vl (AndroidControl only) |
Do NOT modify the template scripts in-place. Follow this procedure:
read_file to get its full content.clawgui-eval/scripts/generate_scripts/ if it does not exist.clawgui-eval/scripts/generate_scripts/<EXPERIMENT_NAME>.sh.| Variable | What to set |
|---|---|
EXPERIMENT_NAME | User-provided experiment label |
MODEL_PATH | User-provided model path or HuggingFace ID |
BENCHMARK | User-provided or derived from dataset + MODEL_TYPE |
NUM_GPUS | From Phase 1 GPU count (or user override) |
Leave all other variables (MAX_TOKENS, TEMPERATURE, TOP_P, TOP_K, MIN_PIXELS, MAX_PIXELS, TV_OR_VT, SYSTEM_PROMPT, USE_CACHE, etc.) at the template defaults unless the user explicitly asks to change them.
Format: <dataset>-<MODEL_TYPE>
Five dataset prefixes:
screenspot-pro-<MODEL_TYPE>screenspot-v2-<MODEL_TYPE>uivision-<MODEL_TYPE>mmbench-gui-<MODEL_TYPE>osworld-g-<MODEL_TYPE>Example: MODEL_TYPE=qwen3vl → BENCHMARK=screenspot-pro-qwen3vl
Same model family with different sizes (e.g. Qwen3 2B/4B/8B) share the same BENCHMARK; differentiate via EXPERIMENT_NAME and MODEL_PATH only.
AndroidControl uses: androidcontrol-high-<MODEL_TYPE> or androidcontrol-low-<MODEL_TYPE>.
Inference is long-running (minutes to hours). The exec tool has a 600s timeout, so use background execution with log redirection.
Step 1a: Get total sample count
cd clawgui-eval && python3 -c "import json; print(len(json.load(open('data/<BENCHMARK>.json'))))"
Save this number as TOTAL_SAMPLES.
Step 1b: Launch the generated script in background
cd clawgui-eval && mkdir -p output/<EXPERIMENT_NAME>/<BENCHMARK> && \
nohup bash scripts/generate_scripts/<EXPERIMENT_NAME>.sh \
> output/<EXPERIMENT_NAME>/<BENCHMARK>/run.log 2>&1 &
echo "PID=$!"
Save the PID.
Step 1c: Poll progress
Periodically check (start at ~30s intervals, increase to ~60s as it stabilizes):
# Is the process still running?
ps -p <PID> -o pid=,stat=,etime= 2>/dev/null || echo "DONE"
# How many samples completed?
wc -l < clawgui-eval/output/<EXPERIMENT_NAME>/<BENCHMARK>/predictions.jsonl 2>/dev/null || echo 0
# GPU utilization
nvidia-smi --query-gpu=index,utilization.gpu,memory.used,memory.total --format=csv,noheader
# Last few log lines
tail -5 clawgui-eval/output/<EXPERIMENT_NAME>/<BENCHMARK>/run.log
Report progress to the user: "Completed X / TOTAL_SAMPLES samples (Y%), GPU utilization Z%".
Step 1d: Detect completion
Inference is done when:
ps -p <PID> returns nothing (process exited), ORpredictions.jsonl line count equals TOTAL_SAMPLESWhen done, check the log tail for errors:
tail -30 clawgui-eval/output/<EXPERIMENT_NAME>/<BENCHMARK>/run.log
If the exit was abnormal (errors in log), report to the user. The --resume flag means re-running will continue from where it stopped.
Judge script templates are at clawgui-eval/scripts/judge/. Pick the one matching the benchmark:
| Template script | Benchmark prefix |
|---|---|
screenspot-pro_run_judge.sh | screenspot-pro |
screenspot-v2_run_judge.sh | screenspot-v2 |
uivision_run_judge.sh | uivision |
mmbench-gui_run_judge.sh | mmbench-gui |
osworld-g_run_judge.sh | osworld-g |
androidcontrol_run_judge.sh | androidcontrol |
Same procedure as inference:
read_file.clawgui-eval/scripts/generate_scripts/<EXPERIMENT_NAME>_judge.sh.EXP_NAME and MODEL_TYPE (the BENCHMARKS array is derived from MODEL_TYPE automatically in most templates).cd clawgui-eval && bash scripts/generate_scripts/<EXPERIMENT_NAME>_judge.sh
Judging is fast (seconds), so run directly with exec -- no need for background execution.
After Judge completes, immediately proceed to Step 3 (Metric). Do NOT stop here.
Metric script templates are at clawgui-eval/scripts/metric/. Pick the one matching the benchmark:
| Template script | Benchmark prefix |
|---|---|
run_metric_screenspot_pro.sh | screenspot-pro |
run_metric_screenspot_v2.sh | screenspot-v2 |
run_metric_uivision.sh | uivision |
run_metric_mmbench_gui.sh | mmbench-gui |
run_metric_osworld_g.sh | osworld-g |
run_metric_androidcontrol.sh | androidcontrol |
Same procedure:
read_file.clawgui-eval/scripts/generate_scripts/<EXPERIMENT_NAME>_metric.sh.EXP_NAME, and BENCHMARK (or MODEL_TYPE if the template derives BENCHMARK from it).cd clawgui-eval && bash scripts/generate_scripts/<EXPERIMENT_NAME>_metric.sh
After all three steps (Infer → Judge → Metric) complete:
output/<EXPERIMENT_NAME>/<BENCHMARK>/metrics.json with read_file and extract the accuracy numbers.run.log to extract inference statistics (throughput, elapsed time).Example format:
=== Evaluation Results ===
Experiment: qwen3vl-2b-screenspot-pro
Model: Qwen/Qwen3-VL-2B-Instruct (qwen3vl)
Benchmark: screenspot-pro-qwen3vl
GPUs: 8
| Metric | Your Result | Official Baseline |
|-----------|-------------|-------------------|
| Accuracy | 44.12% | 48.50% |
Sub-category breakdown:
- Mobile: XX.X%
- Desktop: XX.X%
- Web: XX.X%
Inference: 12.3 samples/s, 103.4s total
If the user ran multiple benchmarks, present a summary comparison table at the end:
| Benchmark | Your Result | Official Baseline |
|-----------------|-------------|-------------------|
| screenspot-pro | XX.X% | XX.X% |
| screenspot-v2 | XX.X% | XX.X% |
| ... | ... | ... |
Use this table to look up official baselines for comparison. - means no official number is available.
| Model | ScreenSpot-Pro | ScreenSpot-V2 | UIVision | MMB-GUI | OSWorld-G |
|---|---|---|---|---|---|
| GUI-G2-7B | 47.50 | 93.30 | - | - | - |
| GUI-Owl 1.5-2B | 57.80 | 89.70 | - | 72.17 | 52.80 |
| GUI-Owl 1.5-4B | 66.80 | 93.20 | - | 83.24 | 63.70 |
| GUI-Owl 1.5-8B | 71.10 | 93.70 | - | 82.52 | 65.80 |
| Qwen3-VL-2B | 48.50 | - | - | - | - |
| Qwen3-VL-4B | 59.50 | - | - | - | - |
| Qwen3-VL-8B | 54.60 | - | - | - | - |
| UI-TARS 1.5-7B | 49.60 | - | - | - | - |
| UI-Venus-7B | 50.80 | 94.10 | 26.50 | - | 58.80 |
| UI-Venus 1.5-2B | 57.70 | 92.80 | 44.80 | 80.30 | 59.40 |
| UI-Venus 1.5-8B | 68.40 | 95.90 | 46.50 | 88.10 | 69.70 |
| MAI-UI-2B | 57.40 | 92.50 | 30.30 | 82.60 | 52.00 |
| MAI-UI-8B | 65.80 | 95.20 | 40.70 | 88.80 | 60.10 |
| StepGUI-4B | 60.00 | 93.60 | - | 84.00 | 66.90 |
When looking up the baseline, match by model name and size. If the exact size is not in the table (e.g. user tested a new size), show the closest available sizes for reference. If no baseline exists (-), note that no official number is available for this combination.
cd clawgui-eval first. All relative paths (data/, image/, output/) depend on this.--benchmark. Without it, main.py runs ALL benchmarks in the registry, which takes extremely long.maiui uses TV_OR_VT=tv, not vt. Do not change this to match other models.--resume is on by default. Re-running the same experiment+benchmark continues from where it stopped rather than starting over.