George Hotz ca77d6cd72 bfloat16 in LLVM (enough for llama 2) (#1293) ...

* add bf16 support to LLVM

* bf16 read works

2023-07-19 20:18:32 -07:00

..

codegen

bfloat16 in LLVM (enough for llama 2) (#1293)

2023-07-19 20:18:32 -07:00

nn

add contiguous (#1246)

2023-07-15 08:36:34 -07:00

runtime

bfloat16 in LLVM (enough for llama 2) (#1293)

2023-07-19 20:18:32 -07:00

shape

strip whitespace

2023-07-19 19:01:53 -07:00

graph.py

Add WHERE ternary (or trinary?) op (#1196)

2023-07-16 00:31:55 -07:00

helpers.py

bfloat16 in LLVM (enough for llama 2) (#1293)

2023-07-19 20:18:32 -07:00

jit.py

Enable JIT tests for supported devices, skip METAL and WEBGPU (#1265)

2023-07-18 11:40:37 -07:00

lazy.py

bfloat16 in LLVM (enough for llama 2) (#1293)

2023-07-19 20:18:32 -07:00

mlops.py

Add WHERE ternary (or trinary?) op (#1196)

2023-07-16 00:31:55 -07:00

ops.py

dedup kernel args v2 (#1272)

2023-07-18 20:03:42 -07:00

state.py

Webgpu support (#1077)

2023-07-12 12:52:06 -07:00

tensor.py

bfloat16 in LLVM (enough for llama 2) (#1293)

2023-07-19 20:18:32 -07:00