huggingface candle issues

huggingface / candle

Minimalist ML framework for Rust

Apache License 2.0

13.48k stars 724 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

"Illegal Instruction" on Older CPUS

#2140 jett06 opened 44 minutes ago
0
Candle won't use half-gemm from cublas when doing fp16 matmul

#2139 lucasavila00 opened 5 hours ago
6
Add a forward_via_f16 method to the qmatmul op.

#2138 LaurentMazare closed 7 hours ago
0
Add the cuda dequantize f16 kernels.

#2137 LaurentMazare closed 7 hours ago
0
Adding direct-F16 quantization

#2136 EricLBuehler closed 5 hours ago
9
Apply the cast before the scaling.

#2135 LaurentMazare closed 19 hours ago
1
Add a sort function, similar to the PyToch one.

#2134 LaurentMazare closed 19 hours ago
0
Make the dtype configurable for phi.

#2133 LaurentMazare closed 1 day ago
0
Add argsort.

#2132 LaurentMazare closed 1 day ago
0
DriverError(CUDA_ERROR_ILLEGAL_ADDRESS, "an illegal memory access was encountered")

#2131 VakeDomen opened 1 day ago
0
A keyboard malfunction created this.

#2130 danielclough closed 1 day ago
1
`Msg("unknown magic 46554747")` when trying to load tensor from GGUF

#2129 shinmao closed 29 minutes ago
2
Phi-3 implementation seems to be buggy on metal devices

#2128 jorgeantonio21 opened 2 days ago
5
Add Olmo models

#2127 Isotr0py closed 2 days ago
1
error when import onnx of yolo8

#2126 xuexl opened 2 days ago
0
quantized_llama ModelWeights path used in quantized Phi3

#2125 AntBlo closed 2 days ago
2
Bug Fix: When converting a tensor to a variable, clone if the tensor is already a variable.

#2124 jsdt opened 4 days ago
2
Support for Microsoft Phi-3 128k context length

#2123 niutech opened 4 days ago
1
Mention phi-v3 in the readmes.

#2122 LaurentMazare closed 4 days ago
0
chore: fix some typos in comments

#2121 hardlydearly closed 19 hours ago
1
Add the phi-3 model.

#2120 LaurentMazare closed 4 days ago
0
Request: Please add support in examples for the new MS/Phi-3 family

#2119 a-agmon closed 4 days ago
2
Add the phi-v3 quantized model.

#2118 LaurentMazare closed 4 days ago
0
Fix for rustfmt.

#2117 LaurentMazare closed 5 days ago
0
candle-onnx: add operators RandomUniform and Exp

#2116 B1rtek closed 5 days ago
1
No compiler check for operation on different tensor type.

#2115 npuichigo opened 5 days ago
1
Fix sigmoid gradient calculation and move sigmoid into a specialized op

#2114 MilkFather opened 5 days ago
4
Add StorageRef.

#2113 LaurentMazare closed 5 days ago
0
error while loading shared libraries: libnvrtc.so.12

#2112 wangjiawen2013 opened 5 days ago
3
Batch llama prompt

#2111 tbogdala closed 5 days ago
1
M2m100 model

#2110 jason-shen opened 6 days ago
0
Equivalent of torch.nonzero ?

#2109 sbucaille opened 6 days ago
1
Processing text prompts in batches for LLMs

#2108 tbogdala closed 5 days ago
4
Use the faster rms-norm kernel for llama.

#2107 LaurentMazare closed 6 days ago
0
Better time measurement for the llama example.

#2106 LaurentMazare closed 6 days ago
0
Build Failure with candle-kernels: nvidia-smi Not Found in Docker Environment, Even Though It Is Available

#2105 Pox-here closed 6 days ago
4
Update tokenizers requirement from 0.15.0 to 0.19.1

#2104 dependabot[bot] closed 6 days ago
0
Update zip requirement from 0.6.6 to 1.1.1

#2103 dependabot[bot] closed 6 days ago
0
Request: PixelShuffle

#2102 oovm opened 1 week ago
1
fix(onnx): No need to install protoc anymore

#2101 oovm opened 1 week ago
0
Derive clone and debug traits for Moondream model

#2100 santiagomed closed 1 week ago
1
Updated quantized phi model

#2099 LaurentMazare closed 1 week ago
1
Small cleanups to the llama multi-process example.

#2098 LaurentMazare closed 1 week ago
0
Handle multiple dimensions in metal QMM + two fixes.

#2097 LaurentMazare closed 1 week ago
0
Add missing onnx operations

#2096 gabotechs closed 1 week ago
1
Run metal and accelerate features on CI

#2095 tomsanbear closed 1 week ago
1
Use llama v3 by default + add to readme.

#2094 LaurentMazare closed 1 week ago
0
Only download the weights in the main process (and not in the child processes).

#2093 LaurentMazare closed 1 week ago
0
Multiprocess/multi-GPU support for llama 3.

#2092 LaurentMazare closed 1 week ago
0
Fix for gemma MQA.

#2091 LaurentMazare closed 1 week ago
0