Issues
- 7
JIT Produces Bad Output SDXL SplitVanillaCFG
#6803 opened by tobias17 - 5
Redo VALIDHACK
#6426 opened by chenyuxyz - 5
`__setitem__` raises "self operand of augmented assign must be contiguous" even though it is
#6352 opened by obadakhalili - 1
Safetensor save, load, re-save does not work
#6294 opened by PaulGustafson - 3
Add bias option to tinygrad.nn.RMSNorm
#6349 opened by EthanBnntt - 0
- 0
ScheduleCache
#6784 opened by geohot - 0
openpilot compile2 failed if MERGE_VIEW=0
#6496 opened by chenyuxyz - 3
Question on tinygrad flux + MCTS optimization
#6759 opened by anthony0727 - 2
`total_length` value in `fetch` function is wrong if `gunzip` is set to True
#6274 opened by Benetti-Hub - 1
- 2
BEAM Search not working as intended
#6504 opened by peppermintcoding - 1
collapse IDIV/ADD from unrolled arange
#6417 opened by geohot - 1
Finite field arithmetic
#6644 opened by 0xWOLAND - 1
Environment variables for backends are confusing
#6689 opened by haraschax - 0
NOLOCALS=1 causes unreadable crash for NV and CUDA
#6687 opened by haraschax - 0
Cat Along a Shard Axis
#6638 opened by tobias17 - 0
qcom vs gpu differents
#6628 opened by nimlgen - 0
`test_broadcastdot` is flaky on 0.9.2
#6542 opened by GaetanLepage - 1
more benchmark on speed frontier
#6539 opened by chenyuxyz - 0
stable diffusion example has broken clip_tokenizer url
#6513 opened by eyaler - 9
The Spiral language showcase
#6510 opened by mrakgr - 4
NV=1 Fails on 3090
#6089 opened by alifuk - 5
Add dtypes.void for UOps
#6045 opened by Qazalin - 0
Delete forced_realize
#6419 opened by zachlite - 0
Bad ordering in allreduce
#6411 opened by geohot - 0
MLX Convs are faster in Resnet18
#6408 opened by geohot - 1
Support `reduction` param in other loss functions
#6247 opened by tim-becker - 1
- 1
NV=1 page fault on 3060
#6229 opened by sixEyees - 3
test_llama_embedding_opt fails on M1 MacBook Pro
#6321 opened by shagler - 2
`pip` 25 will break `pip install -e .`
#6072 opened by roelofvandijk - 0
Why we need runtime ops?
#6221 opened by janetat - 0
.backward() API pre-requisite changes to support higher order derivatives
#6164 opened by davidgonmar - 1
Suggested code improvement
#6141 opened by yonatan895 - 0
why are we assigning a ret variable here?
#6099 opened by MadcowD - 3
BEAM Search not working on metal from fresh install. `no attribute 'commandBuffer'`
#6042 opened by BreakTheBeta - 6
Readonly database error on HPC servers
#5807 opened by ethanthoma - 1
GRAPH=1 Does Not Work On Windows
#6034 opened by QuentinWach - 0
Developing a dockerfile for tinygrad
#6010 opened by Meiye-lj - 1
Fail to run examples with DISK=1
#5769 opened by junaire - 1
- 0
multicore support in CLANG backend
#5994 opened by pasanp-accelr - 0
- 1
Has this software already run on MI250X or MI300X/A ?
#5977 opened by etiennemlb - 1
Release 0.9.2 to Fix the Attribution Error Due to BatchNorm in Beautiful MNIST Example in 0.9.1
#5972 opened by QuentinWach - 1
PF8 Supporton Nvidia
#5929 opened by SaintMil0 - 4
Proposal: supporting second derivatives
#5858 opened by davidgonmar - 0
- 1
Training with MCTS Optimized Kernels
#5809 opened by nanowell