Changes / Feeds / amd/DeepSeek-R1-MXFP4
amd/DeepSeek-R1-MXFP4
3 changes to a tracked key since 2025-05-21, newest first. Head as last read: a7e208e3b71e7cc25f8171d50def2f2008df8f31.
GPU memory to fine-tune it, card by card: DeepSeek-R1-MXFP4 on /m/.
| Date | Advisory | What changed, and what it does |
|---|---|---|
| 2026-02-27 | x_NXMC-2026-0001 breaking | quantization_config changed; the stored format of the weights is different (2026-02-27).359d1ee20ef4 -> d31023d90e8d |
| 2025-10-04 | x_NXMC-2025-0007 breaking | quantization_config changed; the stored format of the weights is different (2025-10-04).a8b84c8ab788 -> 359d1ee20ef4 |
| 2025-06-12 | x_NXMC-2025-0006 | tokenizer model_max_length 512 to 16384; truncation defaults change (2025-06-12).5e05bd6b1c54 -> a8b84c8ab788 |
Pin it, and let a bot move the pin
from_pretrained("amd/DeepSeek-R1-MXFP4", revision="a7e208e3b71e7cc25f8171d50def2f2008df8f31")
With the Renovate preset, a pull request moves this pin when the head moves, and links back here.
[](https://www.neurarch.com/changes-feeds/models/amd/DeepSeek-R1-MXFP4.html)