---
license: apache-2.0
base_model:
- convaiinnovations/laya
library_name: sima-lmm
tags:
- sima.ai
- modalix
- mla
- edge
- decision-model
- modernbert
---

# Laya decision models, compiled for the SiMa.ai MLSoC Modalix

[Laya](https://github.com/NandhaKishorM/laya) by Convai Innovations is a "System-1" decision
model: an encoder (ModernBERT-large, or mmBERT-base for the multilingual one) with a decision
head that answers a typed question about a piece of text in one forward pass. It picks one of
several options, scores on a scale, or says yes or no.

These are those checkpoints compiled for the Modalix MLA. Every graph is a single MLA stage,
with no layers on the CPU, and a decision takes about 19 ms at 128 tokens (8 ms for the
multilingual model).

| folder | model | precision | latency per decision, by tokens | same decision as PyTorch | size |
|---|---|---|---|---|---|
| `general` | Laya | BF16 | 128: 19.5 ms, 256: 35.2 ms, 512: 87.5 ms | 100 / 100 | 3.27 GB |
| `general-int8` | Laya INT8 | A_BF16_W_INT8 | 128: 14 ms | 98 / 100 | 0.67 GB |
| `typed-decisions` | Laya typed-decisions | BF16 | 128: 19.4 ms, 512: 87.5 ms, 1024: 282 ms | 99 / 100 | 4.35 GB |
| `multilingual` | Laya multilingual | BF16 | 128: 8.1 ms, 256: 15.3 ms, 512: 33.9 ms, 1024: 116.5 ms | 100 / 100 | 2.61 GB |
| `dino` | Laya-dino | BF16 | 64: 18 ms | 30 / 30 game states | 0.95 GB |

Latency is one decision on a Modalix DevKit, measured inside the runtime. "Same decision" is
against the fp32 PyTorch model on 100 decisions with the 128-token graph. Precision `BF16` is
weights and activations in bfloat16; `A_BF16_W_INT8` keeps bfloat16 activations and stores
the weights as int8.

[CLM v0.1 8B](https://huggingface.co/Contrastive-LM/CLM-v0.1-8B), a second kind of decision
model on Qwen3-8B, is compiled the same way and published at
[dotimothy/sima-clm](https://huggingface.co/dotimothy/sima-clm).

LiquidAI's d1 models, which also read pictures, are published at
[dotimothy/sima-d1](https://huggingface.co/dotimothy/sima-d1), apart from these because their
licence is another (the LFM Open License v1.0).

## What is in a folder

| file | what it is |
|---|---|
| `laya_s<N>_stage1_mla.elf` | the compiled graph for sequences of up to N tokens |
| `token_embeddings.bf16` | the embedding table, looked up on the CPU |
| `act_tail.f32` | the last layer of the act head, run on the CPU |
| `tokenizer.json` | the checkpoint's tokenizer |
| `laya_config.json` | token ids, calibration temperatures, and which graphs there are |

`models.json` lists every folder with file sizes and SHA-256 sums, here and in the
repositories named above: the web app reads it to know what it can download.

## Using them

The files run on a Modalix board with the runtime and web app from
[neat-decision-studio](https://github.com/dotimothy/neat-decision-studio): its Models page downloads a model from
this repository onto the board and loads it on the MLA. By hand:

```bash
hf download dotimothy/sima-laya --include "general/*" --local-dir .
./laya run general --state "We were billed twice for March." \
    --question '{"type": "noul", "instructions": "Is this a billing problem?"}'
```

They are not usable with PyTorch or ONNX Runtime; for that, use the original checkpoints at
[convaiinnovations/laya](https://huggingface.co/convaiinnovations/laya).

## License and attribution

Apache-2.0. The models are Convai Innovations' Laya checkpoints
(https://github.com/NandhaKishorM/laya, Apache-2.0), compiled without retraining. The
exception is `dino`, whose decision head was fine-tuned on top of the English checkpoint for
a browser game. Zero-shot quality is the checkpoints': the port reproduces PyTorch's answers,
including its wrong ones.