offlinedemo 0.1.0__tar.gz

offlinedemo-0.1.0/PKG-INFO

@@ -0,0 +1,12 @@
+Metadata-Version: 2.3
+Name: offlinedemo
+Version: 0.1.0
+Summary: Easy demonstration of models by offlineisbetter.
+Author: offlineisbetter
+Author-email: offlineisbetter <dev@offlineisbetter.com>
+Requires-Python: >=3.13
+Description-Content-Type: text/markdown
+
+# _offlineisbetter_
+
+Inference for _offlineisbetter_ models.

offlinedemo-0.1.0/README.md

@@ -0,0 +1,3 @@
+# _offlineisbetter_
+
+Inference for _offlineisbetter_ models.

offlinedemo-0.1.0/pyproject.toml

@@ -0,0 +1,18 @@
+[project]
+name = "offlinedemo"
+version = "0.1.0"
+description = "Easy demonstration of models by offlineisbetter."
+readme = "README.md"
+requires-python = ">=3.13"
+dependencies = []
+
+[[project.authors]]
+name = "offlineisbetter"
+email = "dev@offlineisbetter.com"
+
+[project.scripts]
+offlinedemo = "offlinedemo:main"
+
+[build-system]
+requires = ["uv_build>=0.12.19,<0.13.0"]
+build-backend = "uv_build"

offlinedemo-0.1.0/pyproject.toml.orig

@@ -0,0 +1,17 @@
+[project]
+name = "offlinedemo"
+version = "0.1.0"
+description = "Easy demonstration of models by offlineisbetter."
+readme = "README.md"
+authors = [
+    { name = "offlineisbetter", email = "dev@offlineisbetter.com" }
+]
+requires-python = ">=3.13"
+dependencies = []
+
+[project.scripts]
+offlinedemo = "offlinedemo:main"
+
+[build-system]
+requires = ["uv_build>=0.12.19,<0.13.0"]
+build-backend = "uv_build"

offlinedemo-0.1.0/src/offlinedemo/__init__.py

@@ -0,0 +1,43 @@
+# offlineisbetter
+# Copyright (c) 2026- offlineisbetter
+
+from pathlib import Path
+import sys
+import time
+
+from transformers import AutoTokenizer
+import onnxruntime as ort
+
+def main():
+    # Get checkpoint name
+    if len(sys.argv) < 2:
+        print("offlinedemo [checkpoint]")
+        return
+    checkpoint = Path(sys.argv[1])
+
+    # Load checkpoint
+    session = ort.InferenceSession(
+        checkpoint / "model.onnx",
+        providers=["CPUExecutionProvider"],
+    )
+    print("Model loaded!")
+    print()
+
+    # Load tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(
+        checkpoint,
+        local_files_only = True,
+    )
+
+    # Ask user for input
+    user_input = input("offlineisbetter >> ")
+    
+    # Tokenize and run inference
+    start = time.perf_counter()
+    tokens = tokenizer(user_input)
+    result = session.run(None, tokens)[0]
+    duration = time.perf_counter() - start
+    
+    # Report to user
+    print(f"LOGITS:  {result}")
+    print(f"LATENCY: {duration*1000:.3f} ms")