Merge pull request #163 from helicalAI/main

Cleanup and remove torchtext and tensorflow dependencies

.github/workflows/main.yml

@@ -20,9 +20,7 @@ jobs:
       - name: Install dependencies
         run: |
             pip install -r requirements-dev.txt
-            # Currently, the self-hosted runner uses a pre-installed version of mamba-ssm (v2.2.2). Using the latest version (v2.2.4) breaks the pip installation.
-            # TODO fix the installation of helical[mamba-ssm], and add it in the GitHub actions (CI/CD pipeline).
-            pip install .
+            pip install .[mamba-ssm]
             
       # First download before tests as they make use of the downloaded files 
       - name: Download all files
@@ -56,7 +54,7 @@ jobs:
       # because jobs may not be run in the same order, we need to install the dependencies again
       - name: Install helical
         run: |
-            pip install .
+            pip install .[mamba-ssm]
         
       # Required to get the data
       - name: Download all files
@@ -119,10 +117,13 @@ jobs:
         run: |
           python examples/fine_tune_models/fine_tune_mamba2_mrna.py ++device="cuda"
 
-      - name: Execute benchmarking
+      - name: Execute Caduceus
         run: |
-          pip install scanorama
-          python examples/run_benchmark.py
+          python examples/run_models/run_caduceus.py
+    
+      - name: Fine-tune Caduceus
+        run: |
+          python examples/fine_tune_models/fine_tune_caduceus.py
 
   notebooks:
     needs: tests
@@ -141,7 +142,7 @@ jobs:
       # because jobs may not be run in the same order, we need to install the dependencies again
       - name: Install helical
         run: |
-            pip install .
+            pip install .[mamba-ssm]
         
       - name: Reduce datasets to speedup checks
         run: |

.github/workflows/release.yml

@@ -19,7 +19,7 @@ jobs:
         uses: actions/setup-python@v5
         with:
           python-version: 3.11.8
-
+      # Do not install mamba-ssm as it is not supported for machines without GPUs
       - name: Install dependencies
         run: |
             pip install .
@@ -38,9 +38,7 @@ jobs:
       - name: Install dependencies
         run: |
             pip install -r requirements-dev.txt
-            # Currently, the self-hosted runner uses a pre-installed version of mamba-ssm (v2.2.2). Using the latest version (v2.2.4) breaks the pip installation.
-            # TODO fix the installation of helical[mamba-ssm], and add it in the GitHub actions (CI/CD pipeline).
-            pip install .
+            pip install .[mamba-ssm]
             
       # First download before tests as they make use of the downloaded files 
       - name: Download all files
@@ -74,7 +72,7 @@ jobs:
       # because jobs may not be run in the same order, we need to install the dependencies again
       - name: Install helical
         run: |
-            pip install .
+            pip install .[mamba-ssm]
         
       # Required to get the data
       - name: Download all files
@@ -137,10 +135,13 @@ jobs:
         run: |
           python examples/fine_tune_models/fine_tune_mamba2_mrna.py ++device="cuda"
 
-      - name: Execute benchmarking
+      - name: Execute Caduceus
         run: |
-          pip install scanorama
-          python examples/run_benchmark.py
+          python examples/run_models/run_caduceus.py
+      
+      - name: Fine-tune Caduceus
+        run: |
+          python examples/fine_tune_models/fine_tune_caduceus.py
 
   notebooks:
     needs: tests
@@ -159,7 +160,7 @@ jobs:
       # because jobs may not be run in the same order, we need to install the dependencies again
       - name: Install helical
         run: |
-            pip install .
+            pip install .[mamba-ssm]
                     
       - name: Reduce datasets to speedup checks
         run: |

.gitignore

@@ -47,7 +47,6 @@ site
 docs/notebooks/
 
 # Keep unit tests data
-!ci/tests/data/valid_nn_head/*
 !ci/tests/data/*
 
 # Hydra

README.md

@@ -94,6 +94,7 @@ apptainer shell --nv --fakeroot singularity/helical/
 
 ### DNA models:
 - [HyenaDNA](https://helical.readthedocs.io/en/latest/model_cards/hyenadna/)
+- [Caduceus](https://helical.readthedocs.io/en/latest/model_cards/caduceus/)
 
 
 ## Demo & Use Cases