Skip to content

Pull requests: apple/axlearn

New pull request New
Clear current search query, filters, and sorts
47 Open 843 Closed
47 Open 843 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Add support to infer -1 in mesh shape
#113 by lyttonhao was merged Oct 12, 2023 Loading…
Similar to a previous change to repeat.py, changes pipeline.py to split stacked summary values.
#125 by ruomingp was merged Oct 16, 2023 Loading…
Updates chex version to >= 0.1.7 to be compatible with jax 0.4.18.
#124 by ruomingp was merged Oct 15, 2023 Loading…
Exclude arm images for booting VM.
#123 by markblee was merged Oct 14, 2023 Loading…
Add query_scale and key_scale to MultiheadAttention.
#122 by apghml was merged Oct 20, 2023 Loading…
1
upgrade transformers version
#121 by gyin94 was closed Oct 14, 2023 Loading…
1
Fix typo in /docs/01-start.md
#120 by malivinayak was merged Oct 15, 2023 Loading…
2
Bump jax/jaxlib to 0.4.18.
#119 by tgunter was merged Oct 14, 2023 Loading…
[3/N] E2E comparison against LLAMA attention.
#118 by zbwglory was merged Oct 13, 2023 Loading…
Add Nested type.
#117 by apghml was merged Oct 12, 2023 Loading…
Make theta configurable in RoPE
#116 by alex8937 was merged Oct 13, 2023 Loading…
Replaces uses of utils.host_to_global_device_array with map_utils.host_local_array_to_global_array.
#115 by ruomingp was closed Oct 13, 2023 Loading…
Adds a pointer to https://github.com/apple/axlearn/blob/main/docs/01-start.md#preparing-the-cli in the error message.
#114 by ruomingp was merged Oct 13, 2023 Loading…
Add fan_axes to ParameterSpec and to create_parameter_specs_recursively().
#101 by apghml was merged Oct 6, 2023 Loading…
Tighten cryptography/pyopenssl pins.
#112 by markblee was merged Oct 11, 2023 Loading…
Add option of whether to apply rotary position embeddings on value
#111 by alex8937 was merged Oct 11, 2023 Loading…
3
Fix RoPE dtype
#110 by alex8937 was merged Oct 11, 2023 Loading…
Make repeat in RepeatedTransformerLayer configurable.
#109 by xianzhidu was merged Oct 11, 2023 Loading…
Decouples bastion from GCP.
#108 by markblee was merged Oct 11, 2023 Loading…
Workaround JAX PJIT bug that was breaking image summaries.
#107 by apghml was merged Oct 11, 2023 Loading…
Splits output summaries of repeat layers
#106 by ruomingp was merged Oct 10, 2023 Loading…
[1/N] Adding Rotary position embedding test against LLAMA implementation
#105 by zbwglory was merged Oct 10, 2023 Loading…
2
Adds SpmdTrainer.Config.mesh_rules and --mesh_selector to override trainer mesh configuration.
#104 by ruomingp was merged Oct 8, 2023 Loading…
Inherit GCP pins in TPU extras.
#103 by markblee was merged Oct 7, 2023 Loading…
fix issue where flax structs were not recursed into by tree_paths
#102 by apghml was merged Oct 6, 2023 Loading…
2
ProTip! Exclude everything labeled bug with -label:bug.