Add Laya request preprocessing

.github/workflows/ci.yml

@@ -26,5 +26,17 @@ jobs:
         run: cargo clippy --workspace --locked --all-targets -- -D warnings
       - name: Test
         run: cargo test --workspace --locked
+      - name: Official Laya packing parity
+        env:
+          LAYA_TOKENIZER: ${{ runner.temp }}/laya-tokenizer.json
+          LAYA_PACKING_ORACLE: ${{ runner.temp }}/laya-packing.json
+        run: |
+          curl --fail --location --retry 3 \
+            https://huggingface.co/convaiinnovations/laya/resolve/55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851/tokenizer/tokenizer.json \
+            --output "$LAYA_TOKENIZER"
+          curl --fail --location --retry 3 \
+            https://raw.githubusercontent.com/linear3735/system1-omni/5e4dd4215c925ebd93bb9ce4097b27bd6375f7c0/recipe/laya/native/packing-golden.json \
+            --output "$LAYA_PACKING_ORACLE"
+          cargo test --locked -p omni-laya --test packing -- --ignored
       - name: Build
         run: cargo build --workspace --release --locked

Cargo.toml

@@ -1,3 +1,3 @@
 [workspace]
-members = ["src/frontend", "src/models/cua_s1/native"]
+members = ["src/frontend", "src/models/cua_s1/native", "src/models/laya"]
 resolver = "3"

README.md

@@ -49,15 +49,15 @@ Implementation code lives under `src/`; recipes and documentation stay at the re
 | [`recipe/`](recipe/) | Model setup instructions, launch commands, configuration examples, and example requests. |
 | [`docs/`](docs/) | Project documentation and architecture assets. |
 
-The frontend and the Cua-S1 native worker are Cargo workspace members. The other model and backend directories currently document planned work; they do not prescribe process boundaries.
+The frontend, Cua-S1 native worker and Laya checkpoint reader are Cargo workspace members. The other model and backend directories currently document planned work; they do not prescribe process boundaries.
 
 ## Supported models
 
 LAYA can run as an external Python worker for text requests; its in-repository model engine is still planned. The Cua-S1 4B 0.2 `text` adapter runs as a Python worker or as a native worker on CUDA:
 
 | Model | Status |
 | --- | --- |
-| LAYA | [External worker](recipe/laya/README.md); model engine planned |
+| LAYA | [External worker](recipe/laya/README.md); [CPU checkpoint reader](src/models/laya/README.md); model execution planned |
 | Cua-S1 4B 0.2 (`text` adapter) | [Python worker](recipe/cua_s1/text.md); [native worker](recipe/cua_s1/native.md), CUDA, run on sm_89 |
 
 CUDA and Metal coverage will be documented per model as implementations are added and validated.

recipe/laya/native/export_weights.py

@@ -0,0 +1,27 @@
+"""CPU reference hashes for each tensor's FP32, FP16 and BF16 conversions."""
+
+import argparse, hashlib, json
+from pathlib import Path
+import torch
+from safetensors import safe_open
+
+p = argparse.ArgumentParser()
+p.add_argument("checkpoint", type=Path)
+p.add_argument("output", type=Path)
+a = p.parse_args()
+torch.set_num_threads(4)
+rows = []
+with safe_open(a.checkpoint / "model.safetensors", framework="pt", device="cpu") as f:
+    for name in f.keys():
+        x = f.get_tensor(name)
+        row = {"name": name, "shape": list(x.shape), "source_dtype": str(x.dtype)}
+        for key, dtype in [
+            ("f32", torch.float32),
+            ("f16", torch.float16),
+            ("bf16", torch.bfloat16),
+        ]:
+            y = x.to(torch.float32).to(dtype).contiguous()
+            row[key] = hashlib.sha256(y.view(torch.uint8).numpy().tobytes()).hexdigest()
+        rows.append(row)
+a.output.write_text(json.dumps(rows, indent=2))
+print("WEIGHT_ORACLE", len(rows), flush=True)

src/models/cua_s1/native/Cargo.toml

@@ -18,7 +18,7 @@ libloading = "0.8"
 memmap2 = "0.9.9"
 safetensors = "0.8.0"
 serde = "1"
-serde_json = { version = "1.0.149", features = ["float_roundtrip", "preserve_order"] }
+serde_json = { version = "1.0.149", features = ["float_roundtrip", "preserve_order", "raw_value"] }
 # the onig regex backend, as in the Python tokenizers wheel
 tokenizers = { version = "=0.22.2", default-features = false, features = ["onig"] }
 tokio = { version = "1.49.0", features = ["macros", "net", "rt-multi-thread", "sync"] }