[BUG] ESM-2 Finetune notebook - Pre-trained Model Checkpoint esm2/nv_8m:2.0 not found

[mrsaso]

BioNeMo Framework Version

2.6.3

Bug Description

When running the notebook example code for downloading pre-trained model checkpoint, the provided esm2/nv_8m:2.0 checkpoint is not found.

I resolved this by using a different checkpoint.

Steps to Reproduce

Run commands in the notebook. No modifications were made to reproduce this error.

Under the Download Pre-trained Model Checkpoint, run the code

from bionemo.core.data.load import load

checkpoint_path = load("esm2/nv_8m:2.0")

Error Messages and Logs

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[22], line 3
      1 from bionemo.core.data.load import load
----> 3 checkpoint_path = load("esm2/nv_8m:2.0")
      4 print(checkpoint_path)

File /usr/local/lib/python3.12/dist-packages/bionemo/core/data/load.py:175, in load(model_or_data_tag, source, resources, cache_dir)
    172     cache_dir = BIONEMO_CACHE_DIR
    174 if model_or_data_tag not in resources:
--> 175     raise ValueError(f"Resource '{model_or_data_tag}' not found.")
    177 if source == "ngc" and resources[model_or_data_tag].ngc is None:
    178     raise ValueError(f"Resource '{model_or_data_tag}' does not have an NGC URL.")

ValueError: Resource 'esm2/nv_8m:2.0' not found.

Docker Image

nvcr.io/nvidia/clara/bionemo-framework:2.6.3

System Information

Environment Details:

• OS: Ubuntu
• CPU: Intel

GPU Details:

• GPU Model: NVIDIA A100
• GPU Memory: 80GB
• CUDA Version: 12.9
• CUDA Driver: 575.57.08

Additional Context

I resolved this by using another checkpoint esm2/650m:2.0.