issues
search
linkedin
/
Liger-Kernel
Efficient Triton Kernels for LLM Training
BSD 2-Clause "Simplified" License
2.89k
stars
138
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Release Liger-Kernel version 0.3.0
#246
qingquansong
closed
4 hours ago
0
Is not compatible with DoRA?
#245
gotzmann
opened
2 days ago
1
Add label smoothing to FLCE and unit tests
#244
Tcc0403
closed
2 days ago
1
Lable smoothing is not applied and tested in flce
#243
Tcc0403
closed
2 days ago
0
Patch Application Relies on Global State and `AutoLigerKernelForCausalLM.from_config` doesn't work properly
#242
lapp0
closed
21 hours ago
7
Reasons for upcasting the logits dtype outside the kernel
#241
yzhangcs
closed
2 days ago
4
SWIFT Trainer Integration
#240
tastelikefeet
closed
3 days ago
0
Support Z Loss in CE
#239
Tcc0403
opened
3 days ago
1
fused_linear_cross_entropy: Move float32 cast into kernel
#238
hansonw
opened
4 days ago
3
Optimize fused_linear_cross_entropy when weight does not require grads
#237
hansonw
closed
4 days ago
0
Benchmarking phi3 on single A100 40gb GPU: unable to reproduce benchmark results
#236
cosmicBboy
opened
4 days ago
3
Compatibility Issue: PEFT and BitsAndBytesConfig with Liger Kernel. Seeking Alternatives for Quantization and LoRA Fine-Tuning.
#235
GianottiGustavo
opened
4 days ago
11
Support Z Loss in CE
#234
Tcc0403
closed
3 days ago
1
Restore monkey patched modules
#232
austin362667
closed
20 hours ago
4
Triton error on AMD GPUs
#231
eminorhan
opened
5 days ago
6
added `reduction` argument in Cross Entropy Kernel
#230
Ankur-singh
closed
5 hours ago
6
Feat: add kl div to readme
#229
S1ro1
closed
6 days ago
0
[Operator] conv2d
#228
AndreSlavescu
opened
6 days ago
1
Torch compiled FLCE is 2x faster than the current FLCE
#227
ByronHsu
opened
6 days ago
12
(fix) fix pyproject.toml
#226
wizyoung
closed
6 days ago
11
added group norm
#225
denti
opened
1 week ago
0
Add license in ack section
#224
ByronHsu
closed
1 week ago
0
Added HF use-case benchmark script
#223
shimizust
closed
1 week ago
0
Elaborate ack section
#222
ByronHsu
closed
1 week ago
0
Elaborate ack section
#221
ByronHsu
closed
1 week ago
0
add repr infomation for layer_norm and rms_norm
#220
wizyoung
closed
6 days ago
2
Fix compatibility issue on triton=2.3.1
#219
Tcc0403
closed
1 week ago
0
(fix) fix pyproject.toml
#218
wizyoung
closed
1 week ago
1
Update swiglu and geglu forward: zeros_like -> empty_like
#217
IvanYashchuk
closed
6 days ago
9
Reference Unsloth in header
#216
momochen
closed
1 week ago
4
LayerNorm error: TypeError: missing a required argument: 'num_warps'
#215
wizyoung
closed
1 week ago
6
Add support for jamba model with Liger Kernel
#214
yubofredwang
opened
1 week ago
2
minor refactor of rms and layernorm
#213
lancerts
closed
1 week ago
0
Refactor/benchmarking visualizer
#212
S1ro1
closed
6 days ago
4
Refactor: Benchmark visualizer
#211
S1ro1
closed
6 days ago
0
Uplift kernel APIs to top level
#210
austin362667
closed
1 week ago
0
Add `--decimal` flag to benchmarking
#209
austin362667
closed
1 week ago
2
Support Yi-Coder
#208
ryankert01
opened
1 week ago
6
Update layer_norm.py
#207
lancerts
closed
1 week ago
0
MoE kernel
#206
ByronHsu
opened
1 week ago
4
Uplift kernel api to top level
#205
ByronHsu
closed
1 week ago
1
Documentation improvement
#204
merryHunter
closed
1 week ago
4
Update test_rms_norm.py
#203
lancerts
closed
1 week ago
0
ci fix
#202
AndreSlavescu
closed
1 week ago
0
Update the casting logic of RMSNorm
#201
lancerts
closed
1 week ago
2
Ci
#200
AndreSlavescu
closed
1 week ago
0
Support for patching post-model initialization
#199
shimizust
closed
8 hours ago
1
Add label smoothing for cross entropy
#198
Tcc0403
closed
1 week ago
9
Z Loss in CE
#197
Fr0do
opened
1 week ago
2
Refactored benchmark tests
#196
shimizust
closed
1 week ago
3
Next