From c13cb7c8537c16434ecdad2373459c7fedd115b3 Mon Sep 17 00:00:00 2001
From: VED <146507396+ved1beta@users.noreply.github.com>
Date: Fri, 20 Mar 2026 14:53:42 +0530
Subject: [PATCH] feat: add nemotron config (#3506)

* nemotron config exp

* Update examples/nemotron/nemotron-mini-4b-qlora.yaml

Co-authored-by: NanoCode012 <kevinvong@rocketmail.com>

---------

Co-authored-by: NanoCode012 <kevinvong@rocketmail.com>
---
 examples/nemotron/nemotron-mini-4b-qlora.yaml | 57 +++++++++++++++++++
 src/axolotl/monkeypatch/multipack.py          |  1 +
 2 files changed, 58 insertions(+)
 create mode 100644 examples/nemotron/nemotron-mini-4b-qlora.yaml

diff --git a/examples/nemotron/nemotron-mini-4b-qlora.yaml b/examples/nemotron/nemotron-mini-4b-qlora.yaml
new file mode 100644
index 000000000..e796c149c
--- /dev/null
+++ b/examples/nemotron/nemotron-mini-4b-qlora.yaml
@@ -0,0 +1,57 @@
+base_model: nvidia/Nemotron-Mini-4B-Instruct
+
+load_in_8bit: false
+load_in_4bit: true
+
+datasets:
+  - path: fozziethebeat/alpaca_messages_2k_test
+    type: chat_template
+
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.1
+output_dir: ./outputs/nemotron-mini-4b-qlora
+
+adapter: qlora
+lora_model_dir:
+
+sequence_len: 4096
+sample_packing: true
+
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+lora_target_modules:
+  - q_proj
+  - k_proj
+  - v_proj
+  - o_proj
+  - up_proj
+  - down_proj
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 1
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+bf16: auto
+tf32: false
+
+gradient_checkpointing: true
+resume_from_checkpoint:
+logging_steps: 1
+flash_attention: true
+
+warmup_ratio: 0.1
+evals_per_epoch: 1
+saves_per_epoch: 1
+
+special_tokens:
diff --git a/src/axolotl/monkeypatch/multipack.py b/src/axolotl/monkeypatch/multipack.py
index 9e48e73eb..78087acbc 100644
--- a/src/axolotl/monkeypatch/multipack.py
+++ b/src/axolotl/monkeypatch/multipack.py
@@ -59,6 +59,7 @@ SUPPORTED_MULTIPACK_MODEL_TYPES = [
     "ministral3",
     "mistral4",
     "afmoe",
+    "nemotron",
 ]