|
1 | 1 | # Release Notes
|
2 | 2 |
|
3 |
| -## BioNeMo Framework v2.7rc1 |
| 3 | +## BioNeMo Framework v2.7 |
4 | 4 |
|
5 | 5 | ### Updates & Improvements
|
6 | 6 |
|
7 |
| -- Adds a header to SCDL archives, providing improved provenance tracking and supporting future releases. Also adds tracking of the AnnData API coverage in SCDL tests. |
8 |
| - This header stores metadata about the archive and its composite arrays, including a version, the array lengths and data types, and information about the RowFeatureIndexes. This adds the features necessary to fix https://github.com/NVIDIA/bionemo-framework/issues/999 as well as implement simple bit-packing of the rowptr, colptr, and data arrays. It also should make SCDL more secure, enable strict compatibility checking, and open the door to more performance improvements. https://github.com/NVIDIA/bionemo-framework/pull/1030 |
| 7 | +- Evo2 model improvements: |
9 | 8 |
|
10 |
| -- Spike-no-more support for potentially improved training stability: https://github.com/NVIDIA/bionemo-framework/pull/1011 |
| 9 | + - LoRA fine-tuning by @gabenavarro: https://github.com/NVIDIA/bionemo-framework/pull/980. Note: internal CI coverage of LoRA convergence is still a work in progress; therefore, we cannot guarantee convergence. |
| 10 | + - Fix a 2x memory-usage issue during Evo2 generation: https://github.com/NVIDIA/NeMo/pull/14515 |
| 11 | + - Add flash-decode support in inference: https://github.com/NVIDIA/bionemo-framework/pull/1000 |
| 12 | + - Update Rotary Embedding and sequence-length defaults to address incorrect checkpoint conversion: https://github.com/NVIDIA/NeMo/pull/14514 |
| 13 | + - Improvements to tag masking in the Evo2 loss: https://github.com/NVIDIA/bionemo-framework/pull/1008 |
| 14 | + - Support for [Spike-no-more](https://arxiv.org/abs/2312.16903) to improve training stability: https://github.com/NVIDIA/bionemo-framework/pull/1011 |
11 | 15 |
|
12 |
| -- Improvements in tag masking in Evo2 loss https://github.com/NVIDIA/bionemo-framework/pull/1008 |
| 16 | +- Added a header to SCDL archives, providing improved provenance tracking and supporting future releases. It also adds tracking of AnnData API coverage in SCDL tests. |
| 17 | + This header stores metadata about the archive and its composite arrays, including a version; the array lengths and data types; and information about the RowFeatureIndexes. This adds the features necessary to fix https://github.com/NVIDIA/bionemo-framework/issues/999 as well as to implement simple bit-packing of the rowptr, colptr, and data arrays. It should also make SCDL more secure, enable strict compatibility checking, and open the door to further performance improvements: https://github.com/NVIDIA/bionemo-framework/pull/1030 |
13 | 18 |
|
14 |
| -- Flash decode support in inference https://github.com/NVIDIA/bionemo-framework/pull/1000 |
15 |
| - |
16 |
| -- Evo2 Lora feature by @gabenavarro https://github.com/NVIDIA/bionemo-framework/pull/980 |
17 |
| - |
18 |
| -- Added partial-conv benchmark for Evo 2 finetune (from checkpoint) and Evo 2 LoRA finetuning https://github.com/NVIDIA/bionemo-framework/pull/1028 |
| 19 | +- `bionemo-geometric` has been deprecated and removed. The molecular-featurization tooling in this package has moved to [cuik-molmaker](https://github.com/NVIDIA-Digital-Bio/cuik-molmaker). |
19 | 20 |
|
20 | 21 | ### Known Issues
|
21 | 22 |
|
22 |
| -- Users have reported issues in the predict script at sequence lengths over 8192 https://github.com/NVIDIA/bionemo-framework/issues/910 and https://github.com/NVIDIA/bionemo-framework/issues/1048 for example. |
23 |
| -- PR in progress to update Rotary Embedding and sequence length defaults to support incorrect checkpoint conversion https://github.com/NVIDIA/NeMo/pull/14514 |
24 |
| -- PR in progress to fix 2x memory usage issue during generation https://github.com/NVIDIA/NeMo/pull/14515 |
| 23 | +- Users have reported issues with the Evo2 predict script at sequence lengths over 8192; see https://github.com/NVIDIA/bionemo-framework/issues/910 and https://github.com/NVIDIA/bionemo-framework/issues/1048. |
| 24 | +- We have removed `libtiff` from the container due to a known vulnerability, [CVE-2025-9900](https://ubuntu.com/security/CVE-2025-9900). `libtiff` isn't directly used in any BioNeMo code; however, users might face issues with e.g. Pillow or other common image-manipulation libraries inside this container. |
25 | 25 |
|
26 | 26 | ## BioNeMo Framework v2.6.3
|
27 | 27 |
|
|
0 commit comments