From 63f5ad5c52b2727ea6bfc0d328301e5a9cf862b1 Mon Sep 17 00:00:00 2001 From: ruit Date: Sun, 1 Feb 2026 18:28:45 -0800 Subject: [PATCH 1/3] add nanov3 recipe to doc Signed-off-by: ruit --- docs/guides/sft.md | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/guides/sft.md b/docs/guides/sft.md index 2d74914ca2..10cf13fb99 100644 --- a/docs/guides/sft.md +++ b/docs/guides/sft.md @@ -234,6 +234,7 @@ policy: ```bash uv run examples/run_sft.py policy.dtensor_cfg.lora_cfg.enabled=true ``` +For a full configuration example, see the reference recipe: [sft-nanov3-30BA3B-2n4g-fsdp2-lora.yaml](../../examples/configs/recipes/llm/sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml). ### Megatron Configuration Parameters From 8226d273f66c5c12dd2da92ff63790c94c764b2f Mon Sep 17 00:00:00 2001 From: ruit Date: Sun, 1 Feb 2026 22:11:02 -0800 Subject: [PATCH 2/3] add highlight for nano v3 Signed-off-by: ruit --- docs/guides/sft.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/guides/sft.md b/docs/guides/sft.md index 10cf13fb99..230ad82613 100644 --- a/docs/guides/sft.md +++ b/docs/guides/sft.md @@ -191,7 +191,7 @@ Upon completion of the training process, you can refer to our [evaluation guide] ## LoRA Configuration -NeMo RL supports LoRA (Low-Rank Adaptation) for parameter-efficient fine-tuning. LoRA reduces trainable parameters by using low-rank matrices for weight updates while keeping the base model frozen. +NeMo RL supports LoRA (Low-Rank Adaptation) for parameter-efficient fine-tuning, including Nano‑v3 models. LoRA reduces trainable parameters by using low-rank matrices for weight updates while keeping the base model frozen. Notes: - LoRA is supported with DTensor v2 and Megatron backends. Uses the DTensor backend by default. DTensor v1 does not support LoRA (ensure `policy.dtensor_cfg._v2=true` when using DTensor). @@ -234,7 +234,7 @@ policy: ```bash uv run examples/run_sft.py policy.dtensor_cfg.lora_cfg.enabled=true ``` -For a full configuration example, see the reference recipe: [sft-nanov3-30BA3B-2n4g-fsdp2-lora.yaml](../../examples/configs/recipes/llm/sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml). +For the Nano‑v3 SFT LoRA recipe, see:[sft-nanov3-30BA3B-2n4g-fsdp2-lora.yaml](../../examples/configs/recipes/llm/sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml). ### Megatron Configuration Parameters From 81e3dc759268a67d27ad311193b4a00c9c629315 Mon Sep 17 00:00:00 2001 From: ruit Date: Sun, 1 Feb 2026 22:48:54 -0800 Subject: [PATCH 3/3] add nit fix Signed-off-by: ruit --- docs/guides/sft.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/guides/sft.md b/docs/guides/sft.md index 230ad82613..8527c3d7c5 100644 --- a/docs/guides/sft.md +++ b/docs/guides/sft.md @@ -234,7 +234,7 @@ policy: ```bash uv run examples/run_sft.py policy.dtensor_cfg.lora_cfg.enabled=true ``` -For the Nano‑v3 SFT LoRA recipe, see:[sft-nanov3-30BA3B-2n4g-fsdp2-lora.yaml](../../examples/configs/recipes/llm/sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml). +For the Nano‑v3 SFT LoRA recipe, see:[sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml](../../examples/configs/recipes/llm/sft-nanov3-30BA3B-2n8g-fsdp2-lora.yaml). ### Megatron Configuration Parameters