docs: address all open review comments on PR #117

- README.md: use build() directly (from mobius import build) instead of mobius.build()
- execution_providers.md: add 'import onnx_ir as ir' to all code snippets using ir.DataType
- execution_providers.md: fix test file reference _common_test.py -> src/mobius/_build_context_test.py
- ep_quickstart.md: fix wording 'at optimization time' -> 'during build-time validation, before graph construction or optimization starts'
- ep_quickstart.md: fix 'No vendor-specific ops' -> 'No EP-specific vendor fusions (e.g. no GQA/PackQKV)'

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>

Author: justinchuby

README.md

@@ -76,13 +76,15 @@ provider. Pass `execution_provider` to target CUDA, DirectML, WebGPU, and more 
 each with the right set of fused kernels and lowering passes applied automatically:
 
 ```python
+from mobius import build
+
 # CUDA: GQA fusion, SkipLayerNorm, PackQKV
-pkg = mobius.build("meta-llama/Llama-3.2-1B",
-                   execution_provider="cuda", dtype="f16")
+pkg = build("meta-llama/Llama-3.2-1B",
+            execution_provider="cuda", dtype="f16")
 
 # WebGPU: GQA fusion, Shape ops replaced with portable alternatives
-pkg = mobius.build("meta-llama/Llama-3.2-1B",
-                   execution_provider="webgpu", dtype="f16")
+pkg = build("meta-llama/Llama-3.2-1B",
+            execution_provider="webgpu", dtype="f16")
 ```
 
 See the [EP quickstart](docs/ep_quickstart.md) and

docs/ep_quickstart.md

@@ -125,15 +125,16 @@ pkg = mobius.build("meta-llama/Llama-3.2-1B",
                    execution_provider="my-ep", dtype="f16")
 ```
 
-Unrecognised EP names raise `ValueError` at optimization time — safe-fail by default.
+Unrecognised EP names raise `ValueError` during build-time validation, before
+graph construction or optimization starts — safe-fail by default.
 
 ---
 
 ## 5. Common EP configurations at a glance
 
 | Goal | EP | dtype | Notes |
 |---|---|---|---|
-| Portable ONNX (maximum compatibility) | `"default"` | any | No vendor-specific ops |
+| Portable ONNX (maximum compatibility) | `"default"` | any | No EP-specific vendor fusions (e.g. no GQA/PackQKV) |
 | ORT CPU inference | `"cpu"` | `"f32"` | GQA fusion for FP32 |
 | NVIDIA GPU | `"cuda"` | `"f16"` or `"bf16"` | GQA + SkipNorm + PackQKV |
 | Windows GPU (DirectX) | `"dml"` | `"f16"` | RoPE lowered separately |

docs/execution_providers.md

@@ -148,6 +148,7 @@ EpCapabilities(name="trt-rtx", gqa_dtypes={FLOAT16, BFLOAT16},
 Out-of-tree EPs can register at runtime via `register_ep()`:
 
 ```python
+import onnx_ir as ir
 from mobius import EpCapabilities, register_ep
 
 register_ep(EpCapabilities(
@@ -514,6 +515,7 @@ config)`) and want to control EP capabilities without going through
 `build_from_module()`:
 
 ```python
+import onnx_ir as ir
 from mobius import build_context, ep_registry
 
 caps = ep_registry.require("cuda")
@@ -533,7 +535,7 @@ integrating mobius into a custom build pipeline.
 | Build from a custom `nn.Module` | `build_from_module()` |
 | Optimize an existing `ir.Model` | `optimize_model()` |
 | Build graphs with explicit EP context | `build_context()` + `task.build()` |
-| Test EP-conditional component logic | `build_context()` (see `_common_test.py`) |
+| Test EP-conditional component logic | `build_context()` (see `src/mobius/_build_context_test.py`) |
 
 ---
 
@@ -576,6 +578,7 @@ Advanced users who build graphs outside the standard pipeline can set the
 context explicitly:
 
 ```python
+import onnx_ir as ir
 from mobius import build_context, ep_registry
 
 capabilities = ep_registry.require("cuda")
@@ -593,6 +596,7 @@ concurrent builds with different EPs never interfere:
 
 ```python
 import asyncio
+import onnx_ir as ir
 from mobius import build_context, ep_registry
 
 async def build_cuda():