Changes / Feeds / microsoft/Florence-2-large
microsoft/Florence-2-large
3 changes to a tracked key since 2024-06-15, newest first. Head as last read: 21a599d414c4d928c9032694c424fb94458e3594.
GPU memory to fine-tune it, card by card: Florence-2-large on /m/.
| Date | Advisory | What changed, and what it does |
|---|---|---|
| 2024-12-08 | x_NXMC-2024-0013 breaking | derived parameters 204.7M to 207.8M (+3.1M, +0.01 GB of bf16 weights); context window 1024 to 4096 tokens (2024-12-08).6bf179230dd8 -> 00d2f1570b00 |
| 2024-07-20 | x_NXMC-2024-0012 breaking | declared dtype "float32" to "float16" (2024-07-20).865161429344 -> 6bf179230dd8 |
| 2024-06-26 | x_NXMC-2024-0011 breaking | EOS changed from 1 to 2; anything fine-tuned or served against the old value stops at a different token; BOS changed from 2 to 0; prompts built by the old tokenizer start with a different token; pad token 0 to 1; batched training masks a different id (2024-06-26).33ee8e109da7 -> 865161429344 |
Pin it, and let a bot move the pin
from_pretrained("microsoft/Florence-2-large", revision="21a599d414c4d928c9032694c424fb94458e3594")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/microsoft/Florence-2-large.html)