Add extractive alternative section, restore library_name

README.md

@@ -1,18 +1,18 @@
 ---
-base_model: Qwen/Qwen3.5-2B
-datasets:
-- KRLabsOrg/tool-output-extraction-swebench
+license: apache-2.0
 language:
 - en
-license: apache-2.0
-pipeline_tag: text-generation
-library_name: transformers
 tags:
 - code
 - tool-output
 - pruning
 - coding-agents
 - extraction
+datasets:
+- KRLabsOrg/tool-output-extraction-swebench
+base_model: Qwen/Qwen3.5-2B
+pipeline_tag: text-generation
+library_name: transformers
 thumbnail: https://raw.githubusercontent.com/KRLabsOrg/squeez/main/assets/squeez_mascot.png
 ---
 
@@ -116,18 +116,11 @@ messages = [
         "Do not rewrite, summarize, or invent lines."
     )},
     {"role": "user", "content": (
-        "<query>
-Find the failing authentication test
-</query>
-"
-        "<tool_output>
-"
-        "PASSED tests/test_login.py::test_valid_credentials
-"
-        "FAILED tests/test_login.py::test_token_refresh - AssertionError: expected 200 got 401
-"
-        "PASSED tests/test_login.py::test_logout
-"
+        "<query>\nFind the failing authentication test\n</query>\n"
+        "<tool_output>\n"
+        "PASSED tests/test_login.py::test_valid_credentials\n"
+        "FAILED tests/test_login.py::test_token_refresh - AssertionError: expected 200 got 401\n"
+        "PASSED tests/test_login.py::test_logout\n"
         "</tool_output>"
     )},
 ]
@@ -149,8 +142,7 @@ print(response)
 
 **Input** — Chat messages with system prompt:
 - System: extraction instructions (see above)
-- User: `<query>{task}</query>
-<tool_output>{raw_output}</tool_output>`
+- User: `<query>{task}</query>\n<tool_output>{raw_output}</tool_output>`
 
 **Output** — Verbatim lines in XML tags:
 ```
@@ -191,6 +183,19 @@ Examples:
 - cat src/auth/middleware.py | squeez "find the referer validation logic"
 ```
 
+## Smaller extractive alternative
+
+If you don't need a generative model, swap in
+[`KRLabsOrg/verbatim-rag-modern-bert-v2`](https://huggingface.co/KRLabsOrg/verbatim-rag-modern-bert-v2)
+(150M ModernBERT span model). Same CLI:
+
+```bash
+export SQUEEZ_LOCAL_MODEL=KRLabsOrg/verbatim-rag-modern-bert-v2
+pytest -q 2>&1 | squeez "find the failing test"
+```
+
+Head-to-head numbers are on the [verbatim-rag-modern-bert-v2 model card](https://huggingface.co/KRLabsOrg/verbatim-rag-modern-bert-v2).
+
 ## Limitations
 
 - Best on software engineering tool output; not designed for general-purpose summarization
@@ -214,4 +219,4 @@ Apache 2.0
       primaryClass={cs.SE},
       url={https://arxiv.org/abs/2604.04979}, 
 }
-```
+```