OpenThai-SystemOne — mlx-8bit
OpenThai-SystemOne is an open Thai + English System One decision model: one forward
pass answers typed questions (choice over up to 255 options, ordinal score, yes/no noul) about a text / JSON state with
calibrated probabilities, no text generation. It is a Qwen3.5-0.8B text tower (Thai continued pre-training) plus a 256-slot
decision head. This repo is a quantization of v0.3 (commit f3709948).
What is quantized: the tower including the token embeddings (MLX quantizes the embedding table too). The 256-slot decision head and the per-type temperatures stay in fp32 (head.safetensors). Quantization therefore only perturbs the hidden state the head reads.
Format: MLX 8-bit affine (group 64) for Apple Silicon (mlx-lm). mlx-lm runs the tower; the included client applies the decision head on the final hidden states. Size: 800 MB.
Measured on a MacBook Pro M3 Max: 4-bit ≈ 19 ms per 3-question Thai decision (the PyTorch model on MPS: ~150 ms).
Usage
pip install mlx-lm torch transformers safetensors pydantic
huggingface-cli download iapp/OpenThai-SystemOne-MLX-8bit --local-dir openthai-mlx
import sys; sys.path.insert(0, "openthai-mlx")
from openthai_systemone.mlx_client import MLXSystemOneClient
c = MLXSystemOneClient("openthai-mlx")
r = c.system_one("ร้านนี้อาหารอร่อยมาก แต่รอนานเกือบชั่วโมง", {"sentiment": {"type": "choice", "instructions": "ความรู้สึก",
"criteria": {"บวก": None, "ลบ": None, "กลาง": None}}})
print(r.answers["sentiment"].choice, r.answers["sentiment"].probabilities)
Accuracy vs the bf16 original (same records, single option order, first 800 per set)
Macro: public 74.2 (original 74.3), Thai 80.0 (original 80.1).
| subset | bf16 original | this | Δ |
|---|---|---|---|
| public 13-subset bench | |||
| aegis2 (noul) | 83.2 | 83.2 | +0.0 |
| boolq (noul) | 79.7 | 79.3 | -0.3 |
| civil_comments (noul) | 79.0 | 79.0 | +0.0 |
| helpsteer2 (score) | 41.6 | 41.6 | +0.0 |
| massive-de-DE (choice) | 88.3 | 88.3 | +0.0 |
| massive-en-US (choice) | 88.3 | 88.3 | +0.0 |
| multinli (choice) | 89.0 | 88.6 | -0.3 |
| paws (noul) | 94.0 | 94.0 | +0.0 |
| pubmedqa (choice) | 64.0 | 64.0 | +0.0 |
| squad2 (noul) | 89.3 | 89.3 | +0.0 |
| summeval-consistency (score) | 75.0 | 75.7 | +0.7 |
| summeval-relevance (score) | 21.7 | 21.2 | -0.4 |
| vitaminc-dev (choice) | 72.5 | 72.0 | -0.5 |
| macro, public 13-subset bench | 74.3 | 74.2 | -0.1 |
| Thai held-out / eval sets | |||
| banking77 (choice) | 59.1 | 59.1 | +0.0 |
| contrastive_th (choice) | 80.7 | 80.7 | +0.0 |
| contrastive_th (noul) | 83.5 | 83.5 | +0.0 |
| contrastive_th (score) | 78.6 | 76.8 | -1.8 |
| massive_th (choice) | 90.6 | 91.0 | +0.4 |
| prachathai (choice) | 98.3 | 98.3 | +0.0 |
| prachathai (noul) | 93.4 | 93.5 | +0.1 |
| sib200_th (choice) | 77.9 | 78.4 | +0.5 |
| wisesight (choice) | 48.9 | 49.0 | +0.1 |
| wongnai (score) | 64.5 | 64.8 | +0.2 |
| xlam_tools (choice) | 99.4 | 99.4 | +0.0 |
| xnli_th (choice) | 79.8 | 79.8 | +0.0 |
| xnli_th (noul) | 86.8 | 86.4 | -0.4 |
| macro, Thai held-out / eval sets | 80.1 | 80.0 | -0.1 |
Notes
- Scores are single-option-order accuracy on the first 800 records of each set (
scripts/06_eval.py --limit 800), the same records for the original and the quantization.scoresubsets report exact level accuracy. - Base model, data, training and the full benchmark tables: iapp/OpenThai-SystemOne.
- License Apache-2.0 (same as the base). Built by iApp Technology / OpenThaiGPT.