How to use from the
Use from the
Transformers library
# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-classification", model="baselabs/groundhog-1")
# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("baselabs/groundhog-1")
model = AutoModelForMultimodalLM.from_pretrained("baselabs/groundhog-1", device_map="auto")
Quick Links

groundhog-1

groundhog-1 is a 26B parameter mixture-of-experts decision model (about 4B active parameters per token) post-trained from Gemma-4-26B-A4B by Base Labs at Baseten. Provide the model a state (text, optionally with images) and a set of questions and it will return a calibrated probability for every option, answering all questions from one shared-prefix prefill, without generating text.

  • Long documents at a small-model price: trained on sequences up to 131,072 tokens and priced like elephant-1.
  • Same training data and recipe family as bison-1, the #1 decision model on the Artificial Analysis Decision Index v2.1.

For the highest accuracy, see the dense 31B sibling bison-1.

πŸ—’οΈ Model Details

Model Parameters Description
Gemma-4-26B-A4B 26B (about 4B active) General-purpose base model
groundhog-1 26B (about 4B active) Post-trained for single-pass, calibrated decisions
bison-1 31B Dense sibling, highest accuracy
  • Inputs: text and images
  • Training: full fine-tune of the language model on text and image decisions; the vision encoder, expert router, norms and embeddings are frozen
  • Attention: block-bidirectional mask on full-attention layers; sliding-window layers stay causal.
  • Max sequence length: 131,072 tokens

πŸƒ How to use

Baseten Model APIs

groundhog-1 is available on Baseten Model APIs as decision-models/baseten-groundhog-latest, at $0.035 per 1M tokens. It uses the same API as mammoth-1 and elephant-1, and the endpoint accepts text and image inputs.

Request format

A request carries one state and any number of questions. Every question is answered in the same call.

questions = {
    "refund": {"type": "noul", "instructions": "Is the customer asking for a refund?"},
    "team": {
        "type": "choice",
        "instructions": "Which team should handle this?",
        "criteria": {"billing": "Charges, refunds, invoices",
                     "technical": "App or site faults",
                     "fraud": "Suspected unauthorised use"},
    },
    "urgency": {
        "type": "score",
        "instructions": "How urgent is this?",
        "criteria": ["Can wait", "Today", "Blocking the customer now"],
    },
}
request = {"state": "I was charged twice this month, please refund one of them.", "questions": questions}
type criteria answer fields
noul: yes or no optional noul: P(yes)
choice: one of named options {name: description} choice, probabilities
score: ordered levels list of levels, lowest first score, probabilities

Requests longer than the trained length limits are refused, not truncated.

πŸ“Š Performance

Long context (our run)

Benchmark groundhog-1
LongMemEval M, 128k tokens 77.4
LongMemEval S, 128k tokens 79.3

groundhog-1 has not yet been scored on the Artificial Analysis Decision Index.

πŸ”§ Training

  • Data: 6.71M decision examples (Mix 12): core decision skills, safety and abstention, judging and expert domains, document and long-context tasks, and public decision data, including public benchmark sets.
  • Objective: cross-entropy plus a Brier calibration term (Ξ» = 1), with a trained "I don't know" option. About one example in seven carries a probability target instead of a single right answer.
  • Recipe: 1 epoch (5,621 steps), peak LR 3.95e-6, no weight decay, 3% warm-up, bf16, sequences up to 131,072 tokens.
  • Hardware: 16 nodes Γ— 8 B200 (128 GPUs).

License

groundhog-1 is fine-tuned by Baseten from Gemma 4 26B A4B IT, which is licensed under the Apache License 2.0. Baseten's modifications are released under the MIT License. Use of the model is subject to both licenses; see LICENSE and NOTICE.

Attribution

Base model: Gemma 4 26B A4B 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.

Downloads last month
46
Safetensors
Model size
26B params
Tensor type
BF16
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for baselabs/groundhog-1

Finetuned
(256)
this model
Quantizations
1 model