GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
KRLabsOrg
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
KRLabsOrg
Model Tree
Qwen/Qwen3.5-2B
Input Modalities
Output Modalities
Supported Functionality
Evaluated on 618 manually curated held-out examples spanning 27 tool types.
| Model | Prec. | Recall | F1 | Compression |
|---|---|---|---|---|
| Squeez-2B | 0.80 | 0.86 | 0.80 | 0.92 |
| Qwen 3.5 35B A3B (zero-shot) | 0.74 | 0.75 | 0.73 | 0.92 |
| Kimi K2 (zero-shot) | 0.61 | 0.53 | 0.68 | 0.94 |
| Qwen 3.5 2B (untrained) | 0.42 | 0.53 | 0.55 | 0.82 |
The fine-tuned 2B model is also the most precise system in the comparison, indicating it has learned a tool-specific extraction policy rather than relying on generic instruction following.
| Pattern | Example | Squeez-2B | Baseline failure |
|---|---|---|---|
| Precise selection | git_log, 21 lines — find one commit | Selects the single correct entry | Qwen 35B picks a plausible but wrong commit |
| Failure-block extraction | Service log, 176 lines — two similar TLS errors | Returns the correct 5-line block | Qwen 35B picks the wrong TLS error (different timestamp) |
| Correct empty prediction | docker_logs, 316 lines — no matching evidence | Returns empty output | Qwen 35B generates "No relevant lines found..." |
On the 59 negative examples in the test set, Squeez-2B correctly returns empty output 80% of the time. Qwen 35B returns empty only 7% of the time.
bash
pip install squeez # With vLLM servervllm serve KRLabsOrg/squeez-2b --dtype bfloat16 --max-model-len 16384export SQUEEZ_SERVER_URL=http://localhost:8000/v1 pytest -q 2>&1 | squeez "find the failure block"git log --oneline -50 | squeez "find the commit that changed CSRF handling"cat src/auth/middleware.py | squeez "find the referer validation logic"python
from squeez.inference.extractor import ToolOutputExtractor # vLLM serverextractor = ToolOutputExtractor(base_url="http://localhost:8000/v1") # Or localextractor = ToolOutputExtractor(model_path="KRLabsOrg/squeez-2b") filtered = extractor.extract( task="Find the failing test block", tool_output=raw_output,)python
from transformers import AutoModelForCausalLM, AutoTokenizerimport torch model_name = "KRLabsOrg/squeez-2b"tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype=torch.bfloat16, device_map="auto", trust_remote_code=True,) messages = [ {"role": "system", "content": ( "You prune verbose tool output for a coding agent. " "Given a focused extraction query and one tool output, return only the " "smallest verbatim evidence block(s) the agent should read next. " "Return the kept text inside <relevant_lines> tags. " "Do not rewrite, summarize, or invent lines." )}, {"role": "user", "content": ( "<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>" )},] prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)inputs = tokenizer(prompt, return_tensors="pt").to(model.device) with torch.no_grad(): outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.1, do_sample=True) response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)print(response)# <relevant_lines># FAILED tests/test_login.py::test_token_refresh - AssertionError: expected 200 got 401# </relevant_lines>Input — Chat messages with system prompt:
<query>{task}</query>\n<tool_output>{raw_output}</tool_output>Output — Verbatim lines in XML tags:
text
<relevant_lines>{only the lines that matter, copied verbatim}</relevant_lines>SWE-bench derived (14): read_file | grep | git_log | git_blame | git_diff | test_output | python | curl | pip_install | ls | lint_output | build_output | type_check | coverage
Synthetic multi-ecosystem (13): npm_build | tsc | npm_install | docker_logs | docker_build | make_cmake | kubectl | cargo_build | go_build | mvn_gradle | terraform | mypy_pyright | eslint
| Base model | Qwen/Qwen3.5-2B |
| Method | LoRA (r=16, alpha=32) via Unsloth |
| Training data | 10,508 examples (SWE-bench + synthetic) |
| Epochs | 3 |
| Max sequence length | 20,000 tokens |
| Learning rate | 2e-4 |
| Batch size | 8 (32 effective with 4x gradient accumulation) |
| Hardware | Single NVIDIA A100 80GB |
Add to your CLAUDE.md or agent system prompt:
text
When you invoke a shell command, pipe it through `squeez` and describe what you need.Examples:- bun test 2>&1 | squeez "did the tests pass?"- git log --oneline -50 | squeez "find the commit that broke CSRF"- cat src/auth/middleware.py | squeez "find the referer validation logic"If you don't need a generative model, swap in
KRLabsOrg/verbatim-rag-modern-bert-v2
(150M ModernBERT span model). Same CLI:
bash
export SQUEEZ_LOCAL_MODEL=KRLabsOrg/verbatim-rag-modern-bert-v2pytest -q 2>&1 | squeez "find the failing test"Head-to-head numbers are on the verbatim-rag-modern-bert-v2 model card.
openai/gpt-oss-120b — may not fully reflect real-world distributions for all ecosystemsApache 2.0
bibtex
@misc{kovács2026squeeztaskconditionedtooloutputpruning, title={Squeez: Task-Conditioned Tool-Output Pruning for Coding Agents}, author={Ádám Kovács}, year={2026}, eprint={2604.04979}, archivePrefix={arXiv}, primaryClass={cs.SE}, url={https://arxiv.org/abs/2604.04979}, }| Adjacent over-selection | Build output, 110 lines — Dockerfile error | Finds the right error + nearby noise | Qwen 35B misses the Dockerfile error entirely |
| Dataset | KRLabsOrg/tool-output-extraction-swebench |
<relevant_lines>{only the lines that matter, copied verbatim}</relevant_lines>When you invoke a shell command, pipe it through `squeez` and describe what you need.Examples:- bun test 2>&1 | squeez "did the tests pass?"- git log --oneline -50 | squeez "find the commit that broke CSRF"- cat src/auth/middleware.py | squeez "find the referer validation logic"pip install squeez # With vLLM servervllm serve KRLabsOrg/squeez-2b --dtype bfloat16 --max-model-len 16384export SQUEEZ_SERVER_URL=http://localhost:8000/v1 pytest -q 2>&1 | squeez "find the failure block"git log --oneline -50 | squeez "find the commit that changed CSRF handling"cat src/auth/middleware.py | squeez "find the referer validation logic"export SQUEEZ_LOCAL_MODEL=KRLabsOrg/verbatim-rag-modern-bert-v2pytest -q 2>&1 | squeez "find the failing test"from squeez.inference.extractor import ToolOutputExtractor # vLLM serverextractor = ToolOutputExtractor(base_url="http://localhost:8000/v1") # Or localextractor = ToolOutputExtractor(model_path="KRLabsOrg/squeez-2b") filtered = extractor.extract( task="Find the failing test block", tool_output=raw_output,)from transformers import AutoModelForCausalLM, AutoTokenizerimport torch model_name = "KRLabsOrg/squeez-2b"tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype=torch.bfloat16, device_map="auto", trust_remote_code=True,) messages = [ {"role": "system", "content": ( "You prune verbose tool output for a coding agent. " "Given a focused extraction query and one tool output, return only the " "smallest verbatim evidence block(s) the agent should read next. " "Return the kept text inside <relevant_lines> tags. " "Do not rewrite, summarize, or invent lines." )}, {"role": "user", "content": ( "<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>" )},] prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)inputs = tokenizer(prompt, return_tensors="pt").to(model.device) with torch.no_grad(): outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.1, do_sample=True) response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)print(response)# <relevant_lines># FAILED tests/test_login.py::test_token_refresh - AssertionError: expected 200 got 401# </relevant_lines>@misc{kovács2026squeeztaskconditionedtooloutputpruning, title={Squeez: Task-Conditioned Tool-Output Pruning for Coding Agents}, author={Ádám Kovács}, year={2026}, eprint={2604.04979}, archivePrefix={arXiv}, primaryClass={cs.SE}, url={https://arxiv.org/abs/2604.04979}, }