---
license: apache-2.0
library_name: laya
pipeline_tag: text-classification
base_model: convaiinnovations/laya
datasets:
  - allenai/WildChat-1M
language: [en, de, fr, es, it, pt, nl, pl, tr, ru, ar, zh, ja, ko]
tags: [routing, llm-routing, model-selection, prompt-difficulty, laya, system-one, calibrated-decisions, multilingual]
---

# Raya — a fast multilingual LLM router (Laya fine-tune)

**Raya picks which model tier should answer a prompt — a small fast model, a mid-tier model, or a
frontier model — in one ~17 ms forward pass, about 20× faster than a hosted routing API.**

- **Robust to how you phrase the routing question.** Raya scores 80–81% on all three question styles we
  tested. On the difficulty-score style it beats TypeSafe Jev by ~10 points
  (**80.3% vs 70.5%**, paired McNemar p < 0.001).
- **Matches or beats Jev on the minimal routing question in Japanese (90 vs 87), Polish (86 vs 83),
  Spanish (87 vs 85), German (74 vs 72), English (86 vs 86), and Arabic (73 vs 73).**
- **Open and self-hostable.** A 300M-parameter model with no per-call cost, served through Laya's
  Jev-compatible HTTP API.

Raya is [Laya](https://huggingface.co/convaiinnovations/laya)'s multilingual decision model
(mmBERT-base encoder), fine-tuned end-to-end for 3-tier routing. It is not affiliated with Convai
Innovations or TypeSafe.

## Quick start

```python
import laya  # pip install laya  (tested with laya 0.3.7 and 0.3.20; identical results)

raya = laya.Agent("TextCortex/raya", device="cuda")   # or "mps" / "cpu"

ROUTE = {
    "type": "choice",
    "instructions": "Route this prompt to a model.",
    "criteria": {
        "small_model": "simple requests",
        "medium_model": "moderately complex requests",
        "frontier_model": "very hard requests",
    },
}

out = raya.system_one({"prompt": "Schreibe eine professionelle E-Mail an einen Kunden …"}, {"route": ROUTE})
print(out["answers"]["route"])
# {'choice': 'medium_model', 'probabilities': {'small_model': 0.24, 'medium_model': 0.74, 'frontier_model': 0.03}, ...}
```

Raya was trained on three routing questions: the minimal choice above, a detailed rubric choice, and a
3-level difficulty score. Use one of those. Option order does not matter because options were shuffled in
training. Raya serves through Laya's Jev-compatible HTTP server (`POST /v1/systemone`).

## Train your own

The code that trained Raya is in [`training/`](training/README.md), generalised so you can train a fast
decision model on your own data:

- **Any choice or score question:** LLM routing like Raya, ticket triage, intent detection, escalation.
- **Labelling included:** labels come from two independent LLM annotators, which is how Raya's labels
  were made.
- **Adapt Raya:** start from Raya to fit it to your own traffic (`--base TextCortex/raya`).
- **Serving:** exports to ONNX for CPU serving, checked against PyTorch.

```bash
pip install -r training/requirements.txt
python training/train.py --task training/task.example.json --data my_labelled.jsonl --out my-router
```

## CPU inference (ONNX)

No GPU? `onnx/` has ONNX Runtime builds of Raya that run through Laya's own `ONNXAgent`, which gives the same answer format as `laya.Agent`. Use a **512-token** input budget, which is what Raya was trained on.

```python
from huggingface_hub import snapshot_download
from laya.onnx_agent import ONNXAgent   # pip install laya onnxruntime

FILE = "onnx/raya.onnx"                 # see the table below
path = snapshot_download("TextCortex/raya", allow_patterns=["rl_agent_config.json", "tokenizer/*", "encoder/*", FILE])
raya = ONNXAgent(path, onnx_path=f"{path}/{FILE}")
raya.cfg["max_len"] = 512
print(raya.system_one({"prompt": "Schreibe eine professionelle E-Mail …"}, {"route": ROUTE})["answers"]["route"])  # ROUTE as above
```

**Which file?** If you're not sure, use `raya.onnx`. At the same token budget it matches the PyTorch model exactly (0 of 563 choices changed, probabilities within 0.0001), on any CPU. On CPUs with VNNI int8 instructions (Intel Cascade Lake or Alder Lake and newer, AMD Zen 4 and newer), `raya-int8-blockwise.onnx` keeps accuracy and was about 10–15% faster on our Intel i9-13900. The other int8 files are faster still, but lose about 2 accuracy points.

| File | Size | Accuracy | Choices changed vs PyTorch | Intel i9-13900, VNNI (p50 / p95) | AMD EPYC 7502P, no VNNI (p50 / p95) |
|---|---|---|---|---|---|
| `raya.onnx` (fp32) | 1.23 GB | **81.2%** | 3 / 563 | 38 / 297 ms | 60 / 487 ms |
| `raya-int8-blockwise.onnx` | 0.89 GB | **81.5%** | 5 / 563 | **34 / 282 ms** | 127 / 975 ms |
| `raya-int8.onnx` | 0.92 GB | 79.0% | 53 / 563 | 24 / 200 ms | not recommended |
| `raya-int8-emb.onnx` | 0.35 GB | 79.4% | 47 / 563 | 24 / 201 ms | not recommended |
| `raya-int8-mixed.onnx` | 1.05 GB | 79.2% | 41 / 563 | 31 / 242 ms | not recommended |
| PyTorch reference (1,024-token budget) | — | 81.0% | — | 42 / 421 ms | 89 / 526 ms |

All rows use the same 563-prompt benchmark below, with the minimal routing question and single requests on 8 threads. The fp32 and block-wise rows include the switch from a 1,024- to a 512-token budget, which accounts for 3 of their changed choices. Per-tensor int8 (the last three files) overflows on CPUs without VNNI; on the EPYC it dropped to 54–80% depending on settings. 16 threads was slower than 8 on both CPUs. `onnx/SHA256SUMS` has checksums and `onnx/benchmark_results.jsonl` has the raw results.

## Benchmark: 3-tier routing on real multilingual prompts

**Test set.** 563 first-turn prompts from WildChat-1M (shards never used for training), 14 languages,
~40 per language, 75% natural traffic + 25% hard-leaning. Gold tier = the tier two independent blind
annotators (Claude Opus and Claude Sonnet, same written rubric) **both** assigned; the annotators agree on
78% of prompts, which is the practical ceiling. Tier mix: 225 small / 317 medium / 21 frontier.

| Question style | Raya | TypeSafe Jev 1.13 | Laya 0.3.7 (stock) | Von 1.1 (stock) |
|---|---|---|---|---|
| Difficulty score (3 levels) | **80.3%** | 70.5% | 61.6% | 42.3% |
| Rubric choice | 81.0% | **84.2%** | 47.1% | 58.8% |
| Minimal choice | 80.8% | **84.5%** | 55.2% | 54.5% |
| **Latency (p50)** | **17 ms** (1× GPU, in-process) | ~330 ms (hosted API) | — | — |

Always answering "medium" scores 56.3%.

**Per language, minimal choice** (accuracy %, n = 31–53 per language):

| | ja | pl | es | de | en | ar | it | ko | nl | ru | fr | zh | pt | tr |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Raya | **90** | **86** | **87** | **74** | 86 | 73 | 87 | 88 | 78 | 82 | 78 | 79 | 76 | 67 |
| Jev | 87 | 83 | 85 | 72 | 86 | 73 | **90** | **91** | **81** | **89** | **86** | **88** | **90** | **82** |

Confusion, minimal choice (rows = gold small/medium/frontier, cols = Raya): `[[169, 56, 0], [30, 279, 8], [0, 14, 7]]`.

**Limitations of these numbers.** On the choice questions Jev is ahead overall by 3–4 points. That gap
is not statistically significant at these sample sizes (p = 0.07 and 0.13), but it is consistent. Raya is
weakest in Turkish, Portuguese, and Chinese. Differences of less than ~10 points within a single language
are within noise. Like Jev, Raya rarely routes to the frontier tier (7 of 21). Gold labels come from LLM
annotators, not from measured downstream answer quality.

## Training

- **Architecture:** Laya decision model — mmBERT-base encoder + 2-layer transformer decision head with
  per-option `[MASK]` scoring, initialised from `convaiinnovations/laya` (`multilingual`) and fine-tuned
  end-to-end.
- **Data (soft labels from two blind annotators; 50/50 target where they disagreed):** WildChat-1M
  first-turn prompts (ODC-BY); synthetic hard / hard-looking prompts across the 14 languages (kept only
  if a blind annotator confirmed the intended tier); additional in-house routing data.
- **Procedure:** soft-target cross-entropy, options shuffled, AdamW (encoder 2e-5, head 1e-4), batch 32,
  sequences capped at 512 tokens, bf16, best epoch chosen by validation accuracy (86.7% on 659 held-out
  validation prompts, minimal choice), per-question temperature fitted on validation. The seed was also
  chosen on validation only.
- **Compute:** one NVIDIA RTX A6000, ~6 minutes.
- **Code:** [`training/`](training/README.md) reproduces this procedure on your own data. The training
  data itself is not published.

## Intended use and limitations

For choosing among 2–3 model tiers for chat/assistant traffic. Not for safety moderation or for more
than 3 tiers. Not validated on multi-turn context. Calibrate a confidence threshold on your own traffic
before auto-escalating.

## Acknowledgements

Laya architecture, runtime, and base checkpoint by Convai Innovations (Apache-2.0). mmBERT by JHU CLSP
(MIT). WildChat-1M by AI2 (ODC-BY). Jev is a product of TypeSafe AI; its scores come from our own runs
via its API (September 2026).