Merge branch 'main' into ad-412

Author: AngeloDanducci

.github/workflows/quality.yml

@@ -44,8 +44,10 @@ jobs:
         run: curl -fsSL https://ollama.com/install.sh | sh
       - name: Start serving ollama
         run: nohup ollama serve &
-      - name: Pull model granite4:micro
-        run: ollama pull granite4:micro
+      - name: Pull models
+        run: |
+          ollama pull granite4:micro
+          ollama pull granite4:micro-h
       - name: Run Tests
         run: uv run -m pytest -v test
       - name: Send failure message tests

docs/alora.md

@@ -37,7 +37,7 @@ Use the `m alora train` command to fine-tune a LoRA or aLoRA adapter requirement
 
 ```bash
 m alora train path/to/data.jsonl \
-  --basemodel ibm-granite/granite-3.2-8b-instruct \
+  --basemodel ibm-granite/granite-4.0-micro \
   --outfile ./checkpoints/alora_adapter \
   --adapter alora \
   --epochs 6 \
@@ -47,6 +47,10 @@ m alora train path/to/data.jsonl \
   --grad-accum 4
 ```
 
+> **Note on Model Selection**: Use non-hybrid models (e.g., `granite-4.0-micro`) for aLoRA training.
+> Hybrid models (`granite-4.0-h-micro`) are recommended for general inference but adapters should be
+> trained on non-hybrid base models for compatibility with the `ibm-granite/rag-intrinsics-lib` repository.
+
 ### 📌 Parameters
 
 | Flag              | Type    | Default   | Description                                      |

docs/dev/requirement_aLoRA_rerouting.md

@@ -37,10 +37,10 @@ from mellea.core import Requirement
 from mellea.backends.adapters import GraniteCommonAdapter
 
 m = start_session(
-    "huggingface.LocalHFBackend:ibm-granite/granite-3.2-8b-instruct")
+    "huggingface.LocalHFBackend:ibm-granite/granite-4.0-micro")
 
 # By default, the AloraRequirement uses a GraniteCommonAdapter with "requirement_check".
-m.backend.add_adapter(GraniteCommonAdapter("ibm-granite/rag-intrinsics-lib", "requirement_check", base_model_name="granite-3.2-8b-instruct"))
+m.backend.add_adapter(GraniteCommonAdapter("ibm-granite/rag-intrinsics-lib", "requirement_check", base_model_name="granite-4.0-micro"))
 
 m.instruct(
     "Corporate wants you to find the difference between these two strings:\n\naaa\naba")

docs/examples/aLora/101_example.py

@@ -13,7 +13,7 @@
 
 # Define a backend and add the constraint aLora
 backend = LocalHFBackend(
-    model_id="ibm-granite/granite-3.2-8b-instruct", cache=SimpleLRUCache(5)
+    model_id="ibm-granite/granite-4.0-h-micro", cache=SimpleLRUCache(5)
 )
 
 custom_stembolt_failure_constraint = HFConstraintAlora(