Merge branch 'main' into dependabot/pip/types-docutils-0.21.0.20240711

CHANGELOG.md

@@ -5,16 +5,19 @@
 ### Breaking Changes
 ...
 
-
 ### Features
 ...
 
 ### Fixes
-- Reverted a bug introduced in `MultipleChunkRetrieverQa` text highlighting.
+...
 
 ### Deprecations 
 ...
 
+## 5.0.3
+
+### Fixes
+- Update docstrings for 'calculate_bleu' in 'BleuGrader' to now correctly reflect float range from 0 to 100 for the return value.
 
 ## 5.0.2
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "intelligence-layer"
-version = "5.0.2"
+version = "5.0.3"
 description = ""
 authors = ["Aleph Alpha Engineering <[email protected]>"]
 readme = "README.md"
@@ -24,7 +24,7 @@ requests = "^2.32.3"
 pycountry = "24.6.1"
 opentelemetry-api = "^1.22.0"
 opentelemetry-sdk = "^1.22.0"
-huggingface-hub = "^0.23.4"
+huggingface-hub = "^0.23.5"
 opentelemetry-exporter-otlp-proto-http = "1.23.0"
 # summary grader
 rouge-score = "^0.1.2"
@@ -36,7 +36,7 @@ argilla = "^1.29.0"
 # lint & format
 mypy = "^1.10.1"
 nbqa = "^1.8.5"
-ruff = "^0.5.1"
+ruff = "^0.5.2"
 pre-commit = "^3.7.1"
 
 # tests
@@ -57,7 +57,7 @@ pandas-stubs = "^2.2.2.240603"
 
 # docs
 sphinx-rtd-theme = "^2.0.0"
-sphinx = "^7.3.7"
+sphinx = "^7.4.5"
 matplotlib = "^3.9.1"
 
 [tool.poetry.group.docs.dependencies]

src/documentation/how_tos/how_to_implement_a_task.ipynb

@@ -32,6 +32,7 @@
     "   1. Pass the Model to the constructor \n",
     "   2. Implement your domain logic in `do_run()`\n",
     "      1. Generate a `Prompt`. Examples for generating prompts are `ControlModel.to_instruct_prompt()`, `PromptTemplate.to_rich_prompt()` or `Prompt.from_text()`\n",
+    "         - We recommend to not write custom `PromptTemplate`s. If necessary, look up the documentation of `PromptTemplate`\n",
     "      2. Run the model with the prompt\n",
     "      3. Map the prompt output to the task output class\n",
     "3. Run and test it"

src/intelligence_layer/evaluation/evaluation/graders.py

@@ -22,7 +22,7 @@ def calculate_bleu(self, hypothesis: str, reference: str) -> float:
             reference: The baseline for the evaluation.
 
         Returns:
-            BLEU-score, float between 0 and 1. Where 1 means perfect match and 0 no overlap.
+            BLEU-score, float between 0 and 100. Where 100 means perfect match and 0 no overlap.
         """
         bleu_score = self.bleu.corpus_score(
             hypotheses=[hypothesis], references=[[reference]]