| --- |
| language: |
| - en |
| license: apache-2.0 |
| library_name: gliclass |
| pipeline_tag: zero-shot-image-classification |
| base_model: knowledgator/gliclass-v4-2b |
| model_name: Apex-core |
| tags: |
| - apex |
| - gliclass |
| - gliclass-v4 |
| - zero-shot-classification |
| - text-classification |
| - multi-label-classification |
| - instruction-following |
| - reranking |
| - multimodal |
| - image-classification |
| --- |
| # Apex-core |
|
|
| **Evidence in, decisions out.** Explore [SCX Apex](https://labs.scx.ai/apex), a family of GLiClass 4 based Large Decision Models. |
|
|
| **A 2B Large Decision Model for everyday text and image decisions on a single GPU.** |
|
|
| Apex-core reads text and images alongside instructions, policies, and candidate answers. Use it for everyday routing, triage, document and screenshot classification, and choosing the next action in your application. |
|
|
| Part of the [SCX Apex family](https://huggingface.co/collections/scx-admin/apex-family-6ad58a4ea2e0c9875ddb4061), **based on GLiClass 4**. Give the model evidence, an instruction, and the answers you would accept. It scores the candidate answers in one forward pass without generating text. |
|
|
| | Property | Value | |
| | --- | --- | |
| | Parameters | Approximately 2B | |
| | Inputs | Text and images | |
| | Backbone | [Qwen/Qwen3.5-2B](https://huggingface.co/Qwen/Qwen3.5-2B) | |
| | Framework | [GLiClass](https://github.com/Knowledgator/GLiClass) | |
| | License | Apache 2.0 | |
| | Scores | Uncalibrated; tune thresholds on representative data | |
|
|
| ## Use an LLM to write. Use Apex to decide. |
|
|
| - **Change the policy, not the model.** Supply instructions, policies, or rubrics at inference time, and change candidate answers as your workflow evolves. |
| - **Set thresholds for your workflow.** This release provides uncalibrated scores, which can cluster near 0 or 1. Evaluate thresholds on representative data before automating decisions or escalating cases to a person or an LLM. |
| - **Make decisions directly.** Get structured label scores for routing, triage, moderation, ranking, extraction decisions, judging, and selecting the next step. |
| - **Bring images and conversations into the decision.** Score text alongside screenshots, receipts, charts, or camera frames. The GLiClass decoder-KV architecture also supports text-only streaming classification with a per-session cache. See the [upstream streaming examples](UPSTREAM_README.md). |
|
|
| ## Choose an Apex model |
|
|
| | Model | Size | Inputs | Best suited to | |
| | --- | --- | --- | --- | |
| | [Apex-flash](https://huggingface.co/scx-admin/Apex-flash) | 0.4B | Text | High-volume decisions on CPU or GPU | |
| | [Apex-core](https://huggingface.co/scx-admin/Apex-core) | 2B | Text and images | Everyday decisions on a single GPU | |
| | [Apex-large](https://huggingface.co/scx-admin/Apex-large) | 9B | Text and images | Knowledge-heavy and visual decisions | |
|
|
| Explore the [Apex family collection](https://huggingface.co/collections/scx-admin/apex-family-6ad58a4ea2e0c9875ddb4061). All three releases provide open weights under Apache 2.0. Apex-core's scores are uncalibrated; evaluate thresholds for each model on your own data. |
|
|
| ## Quick start |
|
|
| Use Python 3.10+ and an up-to-date GLiClass installation: |
|
|
| ```bash |
| pip install -U gliclass |
| ``` |
|
|
| The example uses a GPU with BF16 support. The checkpoint weights occupy approximately 4.7 GB; allow additional memory for inference. |
|
|
| ```python |
| import torch |
| from gliclass import GLiClassModel, ZeroShotClassificationPipeline |
| |
| model_id = "scx-admin/Apex-core" |
| device = "cuda:0" |
| model = GLiClassModel.from_pretrained(model_id, torch_dtype=torch.bfloat16) |
| pipeline = ZeroShotClassificationPipeline( |
| model, |
| model_id, |
| classification_type="multi-label", |
| max_length=8192, |
| device=device, |
| ) |
| |
| results = pipeline( |
| "My parcel has not arrived. Please check the delivery status.", |
| ["delivery issue", "refund request", "technical support"], |
| threshold=0.5, |
| )[0] |
| for result in results: |
| print(result["label"], result["score"]) |
| ``` |
|
|
| Pass `prompt="your instruction or policy"` to guide the decision. Use `classification_type="single-label"` for mutually exclusive choices. Multi-label results may contain more than one label or no labels above the threshold. |
|
|
| ### Images |
|
|
| Use the same pipeline with an image path, URL, or PIL image: |
|
|
| ```python |
| results = pipeline( |
| ["Which fruit is shown?"], |
| ["strawberry", "lime", "fig", "peach"], |
| images=[["fruit.jpg"]], |
| classification_type="single-label", |
| )[0] |
| print(results) |
| ``` |
|
|
| ## Limitations |
|
|
| Predictions depend on label wording, instructions, domain, input length, and thresholds. English is the primary language. The upstream release reports that its scores are uncalibrated: a score of 0.9 does not imply 90% accuracy. Evaluate scores and thresholds on representative data. See the [upstream model card](https://huggingface.co/knowledgator/gliclass-v4-2b) for training details, evaluations, and further limitations. |
|
|
| ## Attribution and license |
|
|
| This release republishes the upstream checkpoint under the Apex name. The weights, configuration, tokenizer, and other inference assets are unchanged; no additional fine-tuning is claimed. |
|
|
| Based on **GLiClass 4**, developed by **Knowledgator**. Original checkpoint: [knowledgator/gliclass-v4-2b](https://huggingface.co/knowledgator/gliclass-v4-2b/tree/40fe023b45e8d97d1ae4f7f9342279ac1c76317d), revision `40fe023b45e8d97d1ae4f7f9342279ac1c76317d`. Distributed under the upstream **Apache License 2.0**; see [LICENSE](LICENSE). The original model card is preserved in [UPSTREAM_README.md](UPSTREAM_README.md) and at the upstream link. |
|
|
| Related work: [GLiClass](https://arxiv.org/abs/2508.07662) and [SCX Router](https://arxiv.org/abs/2609.02292). |
|
|
