Clarify pretrained checkpoint intended use

README.md

@@ -19,9 +19,10 @@ tags:
 
 # bekko-embedding-v1-a8m-pt
 
-bekko-embedding-v1-a8m-pt is the pretrained model used as the base for [bekko-embedding-v1-a8m](https://huggingface.co/hotchpotch/bekko-embedding-v1-a8m). It is the checkpoint before the final retrieval fine-tuning stage and is intended primarily for downstream fine-tuning, pretrain-versus-fine-tuned comparisons, and ablation studies.
+bekko-embedding-v1-a8m-pt is the pretrained model used as the base for [bekko-embedding-v1-a8m](https://huggingface.co/hotchpotch/bekko-embedding-v1-a8m). It is a starting point for additional training on a downstream task, not a model intended for direct use.
 
-For general-purpose retrieval, use the [fine-tuned model](https://huggingface.co/hotchpotch/bekko-embedding-v1-a8m).
+> [!IMPORTANT]
+> Fine-tune this checkpoint for your target task before using it. It has not received the final supervised retrieval fine-tuning and should not be used directly as a general-purpose embedding or retrieval model. For general-purpose retrieval, use [bekko-embedding-v1-a8m](https://huggingface.co/hotchpotch/bekko-embedding-v1-a8m).
 
 ## Model Details
 
@@ -42,22 +43,13 @@ For general-purpose retrieval, use the [fine-tuned model](https://huggingface.co
 
 The model was pretrained with multilingual text pairs using Matryoshka representation learning and quantization-aware training. The 4-layer backbone retains layers 0, 1, 2, and 18 from mmBERT-small.
 
-## Usage
+## Intended Use
 
-```python
-from sentence_transformers import SentenceTransformer
+Use this checkpoint as initialization for further training on a specific downstream task. Appropriate uses include:
 
-model = SentenceTransformer("hotchpotch/bekko-embedding-v1-a8m-pt")
-
-queries = ["What is multilingual retrieval?"]
-documents = ["Multilingual retrieval searches documents across languages."]
-
-query_embeddings = model.encode_query(queries, normalize_embeddings=True)
-document_embeddings = model.encode_document(documents, normalize_embeddings=True)
-scores = query_embeddings @ document_embeddings.T
-```
-
-This pretrained checkpoint has not received the final supervised retrieval fine-tuning used for bekko-embedding-v1-a8m, so its retrieval quality may be lower or less consistent.
+- fine-tuning for retrieval, classification, reranking, or another target task,
+- reproducing or adapting the bekko fine-tuning pipeline,
+- pretrain-versus-fine-tuned comparisons and ablation studies.
 
 ## License