Text Classification
Transformers
Safetensors
English
gemma4
image-text-to-text
decision-model
classification
calibration
routing
guardrails
vision
multimodal
baseten
Instructions to use baselabs/bison-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use baselabs/bison-1 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="baselabs/bison-1")# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("baselabs/bison-1") model = AutoModelForMultimodalLM.from_pretrained("baselabs/bison-1", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Add LICENSE and NOTICE, state licences on the card
Browse files
LICENSE
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
This repository contains a modified version of a third-party model.
|
| 2 |
+
- Baseten's modifications are licensed under the MIT License below.
|
| 3 |
+
- The underlying base model is licensed under the Apache License,
|
| 4 |
+
Version 2.0 (https://www.apache.org/licenses/LICENSE-2.0). Your use of the
|
| 5 |
+
model is also subject to that license. See NOTICE for attribution.
|
| 6 |
+
|
| 7 |
+
MIT License
|
| 8 |
+
|
| 9 |
+
Copyright (c) 2026 Baseten Labs, Inc.
|
| 10 |
+
|
| 11 |
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
| 12 |
+
of this software and associated documentation files (the "Software"), to deal
|
| 13 |
+
in the Software without restriction, including without limitation the rights
|
| 14 |
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
| 15 |
+
copies of the Software, and to permit persons to whom the Software is
|
| 16 |
+
furnished to do so, subject to the following conditions:
|
| 17 |
+
|
| 18 |
+
The above copyright notice and this permission notice shall be included in all
|
| 19 |
+
copies or substantial portions of the Software.
|
| 20 |
+
|
| 21 |
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
| 22 |
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
| 23 |
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
| 24 |
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
| 25 |
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
| 26 |
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
| 27 |
+
SOFTWARE.
|
NOTICE
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
bison-1
|
| 2 |
+
Copyright (c) 2026 Baseten Labs, Inc.
|
| 3 |
+
|
| 4 |
+
The model weights in this repository are a modified version of Gemma 4 31B IT
|
| 5 |
+
(https://huggingface.co/google/gemma-4-31B-it), made available by Google LLC
|
| 6 |
+
under the Apache License, Version 2.0. Baseten modified the weights by further
|
| 7 |
+
training. The Apache License, Version 2.0 is available at
|
| 8 |
+
https://www.apache.org/licenses/LICENSE-2.0. Baseten's modifications are
|
| 9 |
+
licensed under the MIT License (see LICENSE). Gemma is a trademark of Google
|
| 10 |
+
LLC; this model is not affiliated with or endorsed by Google.
|
README.md
CHANGED
|
@@ -1,7 +1,10 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
|
|
|
|
|
|
| 3 |
library_name: transformers
|
| 4 |
base_model: google/gemma-4-31B-it
|
|
|
|
| 5 |
pipeline_tag: text-classification
|
| 6 |
language:
|
| 7 |
- en
|
|
@@ -125,3 +128,11 @@ Scored by Artificial Analysis, November 19, 2026.
|
|
| 125 |
- **Objective**: cross-entropy plus a Brier calibration term (λ = 1), with a trained "I don't know" option. About one example in seven carries a probability target instead of a single right answer.
|
| 126 |
- **Recipe**: 1 epoch (3,747 steps), peak LR 5e-6, no weight decay, 3% warm-up, bf16, sequences up to 131,072 tokens.
|
| 127 |
- **Hardware**: 24 nodes × 8 B200 (192 GPUs).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: other
|
| 3 |
+
license_name: mit-and-apache-2.0
|
| 4 |
+
license_link: LICENSE
|
| 5 |
library_name: transformers
|
| 6 |
base_model: google/gemma-4-31B-it
|
| 7 |
+
base_model_relation: finetune
|
| 8 |
pipeline_tag: text-classification
|
| 9 |
language:
|
| 10 |
- en
|
|
|
|
| 128 |
- **Objective**: cross-entropy plus a Brier calibration term (λ = 1), with a trained "I don't know" option. About one example in seven carries a probability target instead of a single right answer.
|
| 129 |
- **Recipe**: 1 epoch (3,747 steps), peak LR 5e-6, no weight decay, 3% warm-up, bf16, sequences up to 131,072 tokens.
|
| 130 |
- **Hardware**: 24 nodes × 8 B200 (192 GPUs).
|
| 131 |
+
|
| 132 |
+
## License
|
| 133 |
+
|
| 134 |
+
bison-1 is fine-tuned by Baseten from [Gemma 4 31B IT](https://huggingface.co/google/gemma-4-31B-it), which is licensed under the [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0). Baseten's modifications are released under the MIT License. Use of the model is subject to both licenses; see [LICENSE](LICENSE) and [NOTICE](NOTICE).
|
| 135 |
+
|
| 136 |
+
## Attribution
|
| 137 |
+
|
| 138 |
+
Base model: Gemma 4 31B IT, made available by Google LLC, licensed under the Apache License 2.0. Baseten modified the weights by further training. Gemma is a trademark of Google LLC. This model is not affiliated with or endorsed by Google.
|