-
Notifications
You must be signed in to change notification settings - Fork 15
Pull requests: google/tunix
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Change default sharding to
(2, 4)
for GRPO demo.
#111
opened Jul 16, 2025 by
copybara-service
bot
Loading…
Fix pad id and add a unit test for the tokenizer adapter.
#105
opened Jul 14, 2025 by
copybara-service
bot
Loading…
Add the
get_model_input
method to Gemma3 for returning a dummy input
CLA:yes
#101
opened Jul 11, 2025 by
yhtang
Loading…
Add Qwen3 from_pretrained, generation example, and DeepseekV3 model support
#92
opened Jul 3, 2025 by
danikhan632
Loading…
Preparation for fp8 quantized training support
#91
opened Jul 3, 2025 by
copybara-service
bot
Loading…
Change nnx split and store transformer state.
#74
opened Jun 18, 2025 by
copybara-service
bot
Loading…
Load model on the workers CPU to mitigate the resource stress of the client.
#73
opened Jun 12, 2025 by
copybara-service
bot
Loading…
Add an optional section to serve the GRPO-tuned model via vLLM in the GRPO tutorial
#18
opened May 13, 2025 by
copybara-service
bot
Loading…
ProTip!
Updated in the last three days: updated:>2025-07-13.