From 35bf6ea75ad7787a1691605058ae7b3737069de8 Mon Sep 17 00:00:00 2001 From: Botao Chen Date: Thu, 16 Jan 2025 16:31:13 -0800 Subject: [PATCH] Pin torchtune pkg version (#791) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit ## context This is the follow up of https://github.com/meta-llama/llama-stack/pull/674. Since torchtune is still in alpha stage and the apis are not guarantee backward compatible. Pin the torchtune and torchao pkg version to avoid the latest torchtune release breaks llama stack post training. We will bump the version number manually after with the new pkg release some testing ## test ping an old torchtune pkg version (0.4.0) and the 0.4.0 was installed Screenshot 2025-01-16 at 3 06 47 PM --- llama_stack/providers/registry/post_training.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llama_stack/providers/registry/post_training.py b/llama_stack/providers/registry/post_training.py index 3c5d06c05..3bcda6508 100644 --- a/llama_stack/providers/registry/post_training.py +++ b/llama_stack/providers/registry/post_training.py @@ -14,7 +14,7 @@ def available_providers() -> List[ProviderSpec]: InlineProviderSpec( api=Api.post_training, provider_type="inline::torchtune", - pip_packages=["torch", "torchtune", "torchao", "numpy"], + pip_packages=["torch", "torchtune==0.5.0", "torchao==0.8.0", "numpy"], module="llama_stack.providers.inline.post_training.torchtune", config_class="llama_stack.providers.inline.post_training.torchtune.TorchtunePostTrainingConfig", api_dependencies=[