issues
search
hikettei
/
Caten
[wip] Deep Learning Compiler based on Polyhedral Compiler and Light-weight IRs based on Optimizing Pattern Matcher
https://hikettei.github.io/Caten/
Other
20
stars
4
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
reimplementing scheduler.lisp?
#257
hikettei
opened
8 hours ago
0
some refactors on the dynamic shape args determination
#256
hikettei
opened
1 day ago
0
Scheduler: bfs topological sort after scheduling
#255
hikettei
closed
1 day ago
0
SERIALIZE=1 to no fusion
#254
hikettei
closed
1 day ago
0
BugFix: Print `double-float` number in the correct way
#253
elderica
closed
1 day ago
0
wip: refactor recursive-create-groups
#252
hikettei
closed
6 hours ago
0
Workload: Finalizing the scheduler
#251
hikettei
opened
2 days ago
0
insert tpsort-graph after generating vmops
#250
hikettei
closed
2 days ago
0
Load GPT2 Parameters
#249
hikettei
closed
2 days ago
1
BugFix: Fix structures to be visible at compile time from `trivia` library
#248
elderica
closed
3 days ago
1
Enhancement: with-facet, from-facet
#247
hikettei
opened
4 days ago
0
CI: sbcl-bin/latest
#246
hikettei
closed
4 days ago
0
ShapeTracker: tr-stride instead of tr-shape-for-stride + tr-permute
#245
hikettei
closed
2 days ago
1
Feat: with-inference
#244
hikettei
opened
4 days ago
0
Feat: KVCache GPT2
#243
hikettei
closed
4 days ago
0
Hotfix: Simplify the purged args
#242
hikettei
closed
1 day ago
0
Refactor: Memory Planner Newid creates seen
#241
hikettei
closed
4 days ago
0
BugFix: Tensor Shaped Index-Components
#240
hikettei
closed
4 days ago
1
refactor: make sure two groups have the same rank before merging
#239
hikettei
closed
5 days ago
0
Feat: GPT2 Compilation
#238
hikettei
closed
4 days ago
2
Feat: GPT2 Inference Infrastructure (GGUF, StateDict, BPE Tokenizer)
#237
hikettei
closed
5 days ago
1
Enhancement: Support Parallel Compilation
#236
hikettei
closed
6 days ago
1
optimize id->users
#235
hikettei
closed
6 days ago
1
Opt: id->users is zero-cost
#234
hikettei
closed
6 days ago
0
Refactor: Remove the out-of-date envvars
#233
hikettei
closed
6 days ago
1
opt: don't lower the cached schedule-item (3x faster jit compiler)
#232
hikettei
closed
6 days ago
4
O(n) and Fast JIT Compiler Workload (Transformer > 70 layers)
#231
hikettei
opened
6 days ago
1
Support Full Symbolic JIT (Allow duplicated seen for ds)
#230
hikettei
closed
6 days ago
2
Docs: Support English
#229
hikettei
closed
6 days ago
0
Optimize: Dont lower the cached schedule-item
#228
hikettei
closed
6 days ago
0
a lil tweak on MHA+Use UINT64/INT64 in default
#227
hikettei
closed
1 week ago
0
Refactor: MultiHeadAttention
#226
hikettei
closed
1 week ago
1
BugFix: Batch_Norm Scheduling with JIT=1
#225
hikettei
opened
1 week ago
0
Optimize: LayerNorm = 1 Kernels
#224
hikettei
opened
1 week ago
0
Lowerer: BATCH_SIZE=1
#223
hikettei
closed
1 week ago
0
Fix Scheduler Workload
#222
hikettei
closed
1 week ago
0
Enhancement: Beautiful DOT=1, DOT=2
#221
hikettei
opened
1 week ago
0
refactor: expr-index-components
#220
hikettei
closed
1 week ago
0
bugfix: rotate the permutation in schedule.lisp
#219
hikettei
closed
1 week ago
0
merge-views: Reverse masks for NIL
#218
hikettei
closed
1 week ago
0
New Models
#217
hikettei
opened
1 week ago
0
refactor: graph-schedule
#216
hikettei
closed
1 week ago
4
feat: Rotatory Positional Encoding
#215
abourramouss
closed
6 hours ago
6
BugFix: ExprGraph(x = y + y)
#214
hikettei
closed
1 week ago
3
JIT: Fix for :shrink scheduling
#213
hikettei
closed
1 week ago
1
Refactor: Module supports multiple outputs + view ops
#212
hikettei
closed
1 week ago
0
BugFix: Support and test the full dynamic shape compilation (test-dynamic-shape.lisp)
#211
hikettei
closed
6 days ago
1
Workload: Complete MHA Scheduling + Finish !view scheduler
#210
hikettei
closed
1 week ago
0
TODO: Simplify threefry2x32 kernel
#209
abourramouss
opened
1 week ago
0
PY4CL_PYTHON
#208
hikettei
closed
1 week ago
0
Next