ml-explore/mlx

Issues

• [Feature] Add nan_to_num Function

  #1187 opened 4 days ago by abdussamettrkr
  2

• Implement a `torch.rand_like` equivalent?

  #1182 opened 3 days ago by ibehnam
  5

• [Feature] Export Lora Adapters as GGML

  #1186 opened 4 days ago by rmarnold
  0

• [BUG] compile + checkpoint segfaults

  #1162 opened 12 days ago by awni
  1

• gpt-neox

  #1181 opened 5 days ago by aPaleBlueDot
  4

• Potential mx.load() issue with quantized GGUF's

  #1173 opened 6 days ago by rshah918
  1

• [Feature] dlpack device

  #1159 opened 8 days ago by awni
  5

• [BUG] mlx.core.topk throws segmentation fault for large dimension

  #1156 opened 8 days ago by cccyf
  1

• [FEATURE] how to return mlx intermediate layer output similarly to Keras

  #1056 opened a month ago by thegodone
  3

• [Enhancement] be able to override MLX_METAL_VERSION when running cmake

  #1170 opened 10 days ago by davidkoski
  0

• [BUG] Wrong result for sliced matmul on GPU

  #1163 opened 10 days ago by davidkoski
  1

• I'm asking for help with the following error:

  #1166 opened 11 days ago by Daniel-Lee
  1

• [Feature] Expose something like custom VJP in Python

  #1090 opened a month ago by awni
  5

• [BUG] libc++abi crash when using recurrent layer and transformer

  #1063 opened 11 days ago by domschl
  2

• [BUG] JIT compile mode does not work with LoRA

  #1158 opened 11 days ago by awni
  0

• [BUG] Compiled mx.eval(model.state) raises “Attempting to eval an array without a primitive” with mlx.optimizers.Adam

  #1160 opened 13 days ago by igor-pechersky
  4

• Is dlpack supported?

  #1080 opened 23 days ago by justinchuby
  9

• Question about supporting slices of the type a[:, [0]]

  #1148 opened 17 days ago by JoeJoe1313
  2

• Difference in training convergence between PyTorch & MLX

  #1153 opened 16 days ago by derekelewis
  2

• Implement trace analogical to numpy.trace

  #1143 opened 17 days ago by JoeJoe1313
  1

• [BUG] Wrong slice of a 4D array assigned to with GPU

  #1144 opened 17 days ago by jkaercher
  1

• Optimization Plans for Conv2D CPU Execution

  #1130 opened 24 days ago by abdussamettrkr
  3

• problem of using mlx package

  #1146 opened 19 days ago by QianboZang
  2

• [BUG] `np.ndarray` of bfloat16 using ml_dtypes is being interpreted as complex64

  #1075 opened a month ago by lkarthee
  0

• [BUG] mlx gets stuck with high-dimensional array on Linux

  #1134 opened 22 days ago by cccyf
  1

• [BUG] AttributeError in mlx.core.conj and mlx.core.conjugate functions

  #1127 opened 24 days ago by profConradi
  3

• [BUG] Passing `axis=None` into `argpartition` causes `TypeError`

  #1121 opened 24 days ago by hegyibalint
  1

• [BUG] mlx crashes with msg - uncaught exception of type std::invalid_argument: [Scatter::eval_gpu] Does not support int64

  #1076 opened a month ago by lkarthee
  4

• [Feature] something like `mlx.scipy.stats`

  #1107 opened a month ago by PaulScemama
  2

• [BUG] give better diagnostic message when calling compiled code with an eval in it -- currently "Attempting to eval an array without a primitive"

  #1083 opened a month ago by davidkoski
  8

• [Feature] Leak memory on exit

  #1086 opened a month ago by zcbenz
  6

• [Feature] Build flag to make safetensor and GGUF dependencies optional

  #1096 opened a month ago by awni
  0

• [BUG] mx.radians & mx.degrees - unexpected behavior when the input is not an array

  #1094 opened a month ago by Crear12
  0

• [BUG] Matmul gives wrong output for large sizes

  #1051 opened a month ago by awni
  5

• [Feature] arctan2

  #1065 opened a month ago by yrahul3910
  3

• [BUG] Bad result for GPU matmul for specific shape

  #1084 opened a month ago by awni
  1

• Enhancement - Use M series iPads as extra GPU power

  #1089 opened a month ago by petergreis
  1

• [Feature] Rotary Positional Embeddings for Nomic Embed

  #1088 opened a month ago by zanussbaum
  2

• [BUG] arithmetic operations with numpy arrays are not commutative

  #1066 opened a month ago by lkarthee
  3

• [FEATURE REQUEST] mx.grad doesn't alias argnums and argnames

  #1072 opened a month ago by c0g
  5

• 0.12.2 release was not completed

  #1071 opened a month ago by chenrui333
  8

• [BUG] matmul yields different results when using concat

  #1082 opened a month ago by muchi674
  1

• No module named 'mlx.core'; 'mlx' is not a package

  #1069 opened a month ago by hawkheimmer
  2

• [Feature] KANs

  #1068 opened a month ago by Stealeristaken
  1

• [BUG] EOS terminator for mlx_lm generate function

  #1062 opened a month ago by caseybasichis
  1

• [BUG] cannot replicate a keras model into mlx when I reuse keras pretrained weights

  #1057 opened a month ago by thegodone
  0

• [BUG] Unable to install mlx on MacbookPro M3Pro with MacOS 14.4.1

  #1055 opened a month ago by kidoln
  1

• [BUG] broadcast of scalar array in last dimension fails after #1035

  #1052 opened a month ago by davidkoski
  0

• [FEATURE] in keras LayerNorm by default is apply to last dimension only

  #1049 opened a month ago by thegodone
  9

• [BUG] in-place updating of array slice unexpectedly fails due to broadcasting problem

  #1050 opened a month ago by astrotuna201
  2