mawei/tinygrad
1
0
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-06-13 08:28:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

document new envvars (#1676)

Browse Source 
* feat: document some new envvars

* feat: actually put values

* feat: no more cifar torch

* feat: no fakedata
This commit is contained in:
wozeparrot
2023-08-26 20:17:02 -04:00
committed by GitHub
parent 9d93a82354
commit f61d0657d1
1 changed files with 4 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
docs/env_vars.md
View File

@@ -42,7 +42,8 @@ CLANG | [1] | enable Clang backend
LLVM | [1] | enable LLVM backend
LLVMOPT | [1] | enable slightly more expensive LLVM optimizations
LAZY | [1] | enable lazy operations (this is the default)
OPT | [1-4] | optimization level
OPT | [1-3] | optimization level
KOPT | [1-2] | kernel optimization, 1 turns it on, 2 caches the found optimizations
GRAPH | [1] | create a graph of all operations (requires graphviz)
GRAPHPATH | [/path/to] | where to put the generated graph
PRUNEGRAPH | [1] | prune MovementOps and LoadOps from the graph
@@ -130,6 +131,8 @@ Variable | Possible Value(s) | Description
---|---|---
TORCHWEIGHTS | [1] | use torch to initialize weights
DISABLE_BACKWARD | [1] | don't do backward pass
DIST | [1] | enable distributed training
STEPS | [#] | number of steps
### examples/benchmark_train_efficientnet.py & examples/hlb_cifar10.py
@@ -137,13 +140,6 @@ Variable | Possible Value(s) | Description
---|---|---
ADAM | [1] | use the Adam optimizer
### examples/hlb_cifar10.py & xamples/hlb_cifar10_torch.py
Variable | Possible Value(s) | Description
---|---|---
STEPS | [0-10] | number of steps
FAKEDATA | [1] | enable to use random data
### examples/train_efficientnet.py
Variable | Possible Value(s) | Description