📊 View as a histogram

plot

 plot (x:tinygrad.tensor.Tensor, center:str='zero', max_s:int=10000,
       plt0:Any=True, ax:Optional[matplotlib.axes._axes.Axes]=None)

	Type	Default	Details
x	Tensor		Tensor to explore
center	str	zero	Center plot on `zero`, `mean`, or `range`
max_s	int	10000	Draw up to this many samples. =0 to draw all
plt0	typing.Any	True	Take zero values into account
ax	typing.Optional[matplotlib.axes._axes.Axes]	None	Optionally provide a matplotlib axes.
Returns	PlotProxy

Tensor.manual_seed(42)
t = Tensor.randn(100000)+3
plot(t)

plot(t, center="range")

plot(t, center="mean")

plot((t-3).relu())

plot((t-3).relu(), plt0=False)

fig, ax, = plt.subplots(figsize=(6, 2))
fig.tight_layout()
plot(t, ax=ax);

# # |hide
# if torch.cuda.is_available():
#     cudamem = torch.cuda.memory_allocated()
#     print(f"before allocation: {torch.cuda.memory_allocated()}")
#     numbers = torch.zeros((1, 64, 512), device="cuda")
#     torch.cuda.synchronize()
#     print(f"after allocation: {torch.cuda.memory_allocated()}")
#     display(plot(numbers))
#     print(f"after rgb: {torch.cuda.memory_allocated()}")
   
#     del numbers
#     gc.collect()
#     # torch.cuda.memory.empty_cache()
#     # torch.cuda.synchronize()

#     print(f"after cleanup: {torch.cuda.memory_allocated()}")
#     test_eq(cudamem >= torch.cuda.memory_allocated(), True)