439 3 weeks ago

Typed decision model. Recommended file: pplx-decider-27B-Artemis-Q4_K_S.gguf. Use /v1/systemone. Measured Q4_K_S: same answer as BF16 96.8%, 0 confident answers changed on LocalLLaMA/typed-decisions test split, typed decisions.

ollama run turintech/pplx-decider-27b-artemis

Models

View all →

Readme

pplx-decider-27B Artemis

Perplexity’s pplx-decider 27B for typed decisions about text on a 24 GB GPU. It answers typed questions with a probability for each answer, in one pass.

  • 15.28 GB download for the default file, pplx-decider-27B-Artemis-Q4_K_S.gguf.
  • Same answer as the original on 96.8% of 2,000 test decisions, against 95.7% for a stock Q4_K_M we made (16.55 GB); 0 confident answers changed. These decisions also chose the file.
echo "My Stripe integration keeps failing. Please help ASAP." | ollama run turintech/pplx-decider-27b-artemis:15.3gb --ask "Does this message express urgency?"

pplx-decider is Perplexity’s 27B decision model. The Ollama package is text only; for images, use llama-server as the card shows.

From code, call /v1/systemone with a model field:

curl http://localhost:11434/v1/systemone -d '{
  "model": "turintech/pplx-decider-27b-artemis:15.3gb",
  "state": "My Stripe integration keeps failing. Please help ASAP.",
  "questions": {"urgency": {"type": "noul", "instructions": "Does this message express urgency?"}}
}'

Settings

The package sets num_ctx 16384 and num_batch 4096, the card’s settings. The batch size moves the probabilities slightly.

Results

Our default file against a stock Q4_K_M we made with llama-quantize (no public GGUF of this model runs on stock llama.cpp) and the original, on 2,000 typed decisions from the test split. These decisions were also used to choose the file.

File Size Same answer as original Confident answers changed Accuracy
Ours, Q4_K_S 15.28 GB 96.8% 0 70.8%
Stock Q4_K_M, made by us 16.55 GB 95.7% 0 70.7%
BF16 original 53.8 GB reference reference 70.5%

The stock Q4_K_M is larger and closer to the original on the benchmark sample. How it was tested.

Files

Every file we publish for this model. Sizes are download sizes, not memory.

File Size Use it for Ollama tag
pplx-decider-27B-Artemis-Q4_K_S.gguf 15.28 GB Default q4_k_s (latest)
pplx-decider-27B-Artemis-IQ3_M.gguf 14.27 GB Smaller: slightly less memory iq3_m
  • The Ollama tag runs our package, which holds the same file: ollama run turintech/pplx-decider-27b-artemis:<tag>.

Without a tag, -hf and hf.co fetch pplx-decider-27B-Artemis-IQ3_M.gguf, not the default file, so always add the tag.

Limitations

  • On an RTX 5090 this file changed 1 answer the original was more than 80% sure of (0 on the A100), and repeat runs there gave identical answers on 99.6% of decisions.
  • The additional panel historically called held-out has not been verified independent of calibration.
  • The numbers come from llama.cpp, not from Ollama or LM Studio.

Licence

Apache-2.0, the same as perplexity-ai/pplx-decider-v1.1-27b. The package carries the licence and TurinTech’s NOTICE; the Hugging Face repository has both files.