Browse / Agent tools
intel/auto-round
No description on GitHub.
- int4
- quantization
- rounding
- transformers
- vllm
- mxfp4
- nvfp4
- gguf
- sglang
- llms
- vlms
- diffusers
Not tested.Tested on 2026-10-18. The saved checks do not establish a product verdict.
What we checked
Does it run from the README?Not tested.The run
missing prerequisite cannot prove a working operation: {"returncode": 0, "stdout": "", "stderr": "uantizing model.layers.12: 43%|\u2588\u2588\u2588\u2588\u258e | 12/28 [00:03<00:04, 3.68it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.12: 46%|\u2588\u2588\u2588\u2588\u258b | 13/28 [00:03<00:04, 3.72it/s]\rQuantizing model.layers.13: 46%|\u2588\u2588\u2588\u2588\u258b | 13/28 [00:03<00:04, 3.72it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.13: 50%|\u2588\u2588\u2588\u2588\u2588 | 14/28 [00:03<00:03, 3.81it/s]\rQuantizing model.layers.14: 50%|\u2588\u2588\u2588\u2588\u2588 | 14/28 [00:03<00:03, 3.81it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.14: 54%|\u2588\u2588\u2588\u2588\u2588\u258e | 15/28 [00:04<00:03, 3.88it/s]\rQuantizing model.layers.15: 54%|\u2588\u2588\u2588\u2588\u2588\u258e | 15/28 [00:04<00:03, 3.88it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.15: 57%|\u2588\u2588\u2588\u2588\u2588\u258b | 16/28 [00:04<00:03, 3.84it/s]\rQuantizing model.layers.16: 57%|\u2588\u2588\u2588\u2588\u2588\u258b | 16/28 [00:04<00:03, 3.84it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.16: 61%|\u2588\u2588\u2588\u2588\u2588\u2588 | 17/28 [00:04<00:02, 3.74it/s]\rQuantizing model.layers.17: 61%|\u2588\u2588\u2588\u2588\u2588\u2588 | 17/28 [00:04<00:02, 3.74it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.17: 64%|\u2588\u2588\u2588\u2588\u2588\u2588\u258d | 18/28 [00:04<00:02, 3.80it/s]\rQuantizing model.layers.18: 64%|\u2588\u2588\u2588\u2588\u2588\u2588\u258d | 18/28 [00:04<00:02, 3.80it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.18: 68%|\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 19/28 [00:05<00:02, 3.90it/s]\rQuantizing model.layers.19: 68%|\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 19/28 [00:05<00:02, 3.90it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.19: 71%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 20/28 [00:05<00:02, 3.88it/s]\rQuantizing model.layers.20: 71%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 20/28 [00:05<00:02, 3.88it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.20: 75%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 21/28 [00:05<00:01, 3.91it/s]\rQuantizing model.layers.21: 75%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 21/28 [00:05<00:01, 3.91it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.21: 79%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 22/28 [00:05<00:01, 3.94it/s]\rQuantizing model.layers.22: 79%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 22/28 [00:05<00:01, 3.94it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.22: 82%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 23/28 [00:06<00:01, 3.94it/s]\rQuantizing model.layers.23: 82%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 23/28 [00:06<00:01, 3.94it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.23: 86%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 24/28 [00:06<00:01, 3.93it/s]\rQuantizing model.layers.24: 86%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 24/28 [00:06<00:01, 3.93it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.24: 89%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2589 | 25/28 [00:06<00:00, 3.84it/s]\rQuantizing model.layers.25: 89%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2589 | 25/28 [00:06<00:00, 3.84it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.25: 93%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258e| 26/28 [00:06<00:00, 3.87it/s]\rQuantizing model.layers.26: 93%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258e| 26/28 [00:06<00:00, 3.87it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.26: 96%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258b| 27/28 [00:07<00:00, 3.73it/s]\rQuantizing model.layers.27: 96%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258b| 27/28 [00:07<00:00, 3.73it/s]\u001b[38;20m2026-10-18 18:48:20 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.27: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 28/28 [00:07<00:00, 3.79it/s]\u001b[38;20m2026-10-18 18:48:20 INFO shard_writer.py L527: model has been saved to ./tmp_autoround/Qwen3-0.6B-w4g128/\u001b[0m\n\rQuantizing model.layers.27: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 28/28 [00:08<00:00, 3.44it/s]\n\u001b[38;20m2026-10-18 18:48:20 INFO utils.py L426: Immediate saving mode: weights already saved by ShardWriter, saving configs only.\u001b[0m\n\u001b[38;20m2026-10-18 18:48:21 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n", "needs_credentials": true, "needs_browser": false}
Does it do what the README says?No claims established. The saved evidence does not establish an answer.Why
We check claims after it’s running.
Which checks were skipped?1 check skipped.Reasons
Skipped checks count as neither passes nor failures.
Setups: 1 skipped
- Linux / Haiku (2026-10-18): missing prerequisite cannot prove a working operation: {"returncode": 0, "stdout": "", "stderr": "uantizing model.layers.12: 43%|\u2588\u2588\u2588\u2588\u258e | 12/28 [00:03<00:04, 3.68it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.12: 46%|\u2588\u2588\u2588\u2588\u258b | 13/28 [00:03<00:04, 3.72it/s]\rQuantizing model.layers.13: 46%|\u2588\u2588\u2588\u2588\u258b | 13/28 [00:03<00:04, 3.72it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.13: 50%|\u2588\u2588\u2588\u2588\u2588 | 14/28 [00:03<00:03, 3.81it/s]\rQuantizing model.layers.14: 50%|\u2588\u2588\u2588\u2588\u2588 | 14/28 [00:03<00:03, 3.81it/s]\u001b[38;20m2026-10-18 18:48:16 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.14: 54%|\u2588\u2588\u2588\u2588\u2588\u258e | 15/28 [00:04<00:03, 3.88it/s]\rQuantizing model.layers.15: 54%|\u2588\u2588\u2588\u2588\u2588\u258e | 15/28 [00:04<00:03, 3.88it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.15: 57%|\u2588\u2588\u2588\u2588\u2588\u258b | 16/28 [00:04<00:03, 3.84it/s]\rQuantizing model.layers.16: 57%|\u2588\u2588\u2588\u2588\u2588\u258b | 16/28 [00:04<00:03, 3.84it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.16: 61%|\u2588\u2588\u2588\u2588\u2588\u2588 | 17/28 [00:04<00:02, 3.74it/s]\rQuantizing model.layers.17: 61%|\u2588\u2588\u2588\u2588\u2588\u2588 | 17/28 [00:04<00:02, 3.74it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.17: 64%|\u2588\u2588\u2588\u2588\u2588\u2588\u258d | 18/28 [00:04<00:02, 3.80it/s]\rQuantizing model.layers.18: 64%|\u2588\u2588\u2588\u2588\u2588\u2588\u258d | 18/28 [00:04<00:02, 3.80it/s]\u001b[38;20m2026-10-18 18:48:17 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.18: 68%|\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 19/28 [00:05<00:02, 3.90it/s]\rQuantizing model.layers.19: 68%|\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 19/28 [00:05<00:02, 3.90it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.19: 71%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 20/28 [00:05<00:02, 3.88it/s]\rQuantizing model.layers.20: 71%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 20/28 [00:05<00:02, 3.88it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.20: 75%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 21/28 [00:05<00:01, 3.91it/s]\rQuantizing model.layers.21: 75%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 21/28 [00:05<00:01, 3.91it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.21: 79%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 22/28 [00:05<00:01, 3.94it/s]\rQuantizing model.layers.22: 79%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258a | 22/28 [00:05<00:01, 3.94it/s]\u001b[38;20m2026-10-18 18:48:18 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.22: 82%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 23/28 [00:06<00:01, 3.94it/s]\rQuantizing model.layers.23: 82%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258f | 23/28 [00:06<00:01, 3.94it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.23: 86%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 24/28 [00:06<00:01, 3.93it/s]\rQuantizing model.layers.24: 86%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258c | 24/28 [00:06<00:01, 3.93it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.24: 89%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2589 | 25/28 [00:06<00:00, 3.84it/s]\rQuantizing model.layers.25: 89%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2589 | 25/28 [00:06<00:00, 3.84it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.25: 93%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258e| 26/28 [00:06<00:00, 3.87it/s]\rQuantizing model.layers.26: 93%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258e| 26/28 [00:06<00:00, 3.87it/s]\u001b[38;20m2026-10-18 18:48:19 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.26: 96%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258b| 27/28 [00:07<00:00, 3.73it/s]\rQuantizing model.layers.27: 96%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u258b| 27/28 [00:07<00:00, 3.73it/s]\u001b[38;20m2026-10-18 18:48:20 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n\rQuantizing model.layers.27: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 28/28 [00:07<00:00, 3.79it/s]\u001b[38;20m2026-10-18 18:48:20 INFO shard_writer.py L527: model has been saved to ./tmp_autoround/Qwen3-0.6B-w4g128/\u001b[0m\n\rQuantizing model.layers.27: 100%|\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588\u2588| 28/28 [00:08<00:00, 3.44it/s]\n\u001b[38;20m2026-10-18 18:48:20 INFO utils.py L426: Immediate saving mode: weights already saved by ShardWriter, saving configs only.\u001b[0m\n\u001b[38;20m2026-10-18 18:48:21 INFO device.py L1553: 'peak_ram': 3.33GB\u001b[0m\n", "needs_credentials": true, "needs_browser": false}
Needs keys, a login or a browser?Yes.Details
Needs a key or login on some setups
Independent test, not endorsed by the project.