Added MultiHeadSelfAttention.

pyproject.toml

@@ -3,7 +3,23 @@ name = "llmfs"
 version = "0.1.0"
 description = "LLM from scratch"
 readme = "README.md"
-requires-python = ">=3.13"
+requires-python = ">=3.13,<3.15"
 dependencies = [
-    "tiktoken>=0.12.0",
+  "tiktoken>=0.12.0",
+  "torch>=2.11.0",
+  "torchvision>=0.26.0",
 ]
+
+[dependency-groups]
+dev = [
+  "mypy>=1.20.2",
+]
+
+[tool.uv.sources]
+torch = { index = "torch" }
+torchvision = { index = "torch" }
+
+[[tool.uv.index]]
+name = "torch"
+url = "https://download.pytorch.org/whl/cpu"
+explicit = true