diff --git a/.gitmodules b/.gitmodules
index 8d7c7be7e5..ab84439b11 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -11,7 +11,7 @@
 [submodule "3rdparty/Automodel-workspace/Automodel"]
 	path = 3rdparty/Automodel-workspace/Automodel
 	url = https://github.com/NVIDIA-NeMo/Automodel.git
-	branch = yifu/bump-torch-and-hf
+	branch = main
 	shallow = true
 [submodule "3rdparty/Gym-workspace/Gym"]
 	path = 3rdparty/Gym-workspace/Gym
diff --git a/3rdparty/Automodel-workspace/Automodel b/3rdparty/Automodel-workspace/Automodel
index 8b11622f7c..92635e74f4 160000
--- a/3rdparty/Automodel-workspace/Automodel
+++ b/3rdparty/Automodel-workspace/Automodel
@@ -1 +1 @@
-Subproject commit 8b11622f7cbe7462da1541764cbefd89f3c7b0fc
+Subproject commit 92635e74f4fb16784268b9a9fd7b7d6a83fff6c5
diff --git a/3rdparty/Megatron-Bridge-workspace/Megatron-Bridge b/3rdparty/Megatron-Bridge-workspace/Megatron-Bridge
index 45dbf4735e..a2bb70b91b 160000
--- a/3rdparty/Megatron-Bridge-workspace/Megatron-Bridge
+++ b/3rdparty/Megatron-Bridge-workspace/Megatron-Bridge
@@ -1 +1 @@
-Subproject commit 45dbf4735ec3a2034be66ed4183604642a84af0d
+Subproject commit a2bb70b91b827bd6b085a77442c7cf60cfdb59fe
diff --git a/3rdparty/Megatron-Bridge-workspace/setup.py b/3rdparty/Megatron-Bridge-workspace/setup.py
index 38f784707d..a1fa9a77e6 100644
--- a/3rdparty/Megatron-Bridge-workspace/setup.py
+++ b/3rdparty/Megatron-Bridge-workspace/setup.py
@@ -26,9 +26,15 @@
 bridge_package_name = "megatron.bridge"
 
 CACHED_DEPENDENCIES = [
-    "transformers>=5.0.0,<=5.2.0",
+    "transformers>=5.0.0,<=5.3.0",
+    "peft>=0.18.1",
     "datasets>=2.20.0",
     "accelerate",
+    "diffusers>=0.36.0",
+    "peft>=0.18.0",
+    "einops",
+    "imageio",
+    "imageio-ffmpeg",
     "omegaconf>=2.3.0",
     "tensorboard>=2.19.0",
     "typing-extensions",
@@ -44,12 +50,13 @@
     # TODO(https://github.com/NVIDIA-NeMo/RL/issues/2111): upgrade to core_cu13 when we move to CUDA 13 base container
     "transformer-engine[pytorch,core_cu12]",
     "mamba-ssm",
-    "nvidia-resiliency-ext~=0.4.1",
+    "nvidia-resiliency-ext~=0.5.0",
     "causal-conv1d",
     "flash-linear-attention",
     "timm",
     "open-clip-torch>=3.2.0",
     "mlflow>=3.5.0",
+    "comet-ml>=3.50.0",
     "torch>=2.6.0",
 ]
 
diff --git a/3rdparty/Megatron-LM-workspace/Megatron-LM b/3rdparty/Megatron-LM-workspace/Megatron-LM
index 8318b8093e..9e28104173 160000
--- a/3rdparty/Megatron-LM-workspace/Megatron-LM
+++ b/3rdparty/Megatron-LM-workspace/Megatron-LM
@@ -1 +1 @@
-Subproject commit 8318b8093e7d49bce2529a29a14aeb50e5840ee0
+Subproject commit 9e2810417315a7ee93b41d4e234454abd3c16af5
diff --git a/3rdparty/Megatron-LM-workspace/setup.py b/3rdparty/Megatron-LM-workspace/setup.py
index b68b979732..8a7b438bb7 100644
--- a/3rdparty/Megatron-LM-workspace/setup.py
+++ b/3rdparty/Megatron-LM-workspace/setup.py
@@ -49,8 +49,8 @@
     # Dev dependencies from pyproject.toml
     "nvidia-modelopt[torch]; sys_platform != 'darwin'",
     # TODO(https://github.com/NVIDIA-NeMo/RL/issues/2111): upgrade to core_cu13 when we move to CUDA 13 base container
-    "transformer-engine[pytorch,core_cu12]>=2.9.0a0,<2.12.0",
-    "nvidia-resiliency-ext",
+    "transformer-engine[pytorch,core_cu12]",
+    "nvidia-resiliency-ext @ git+https://github.com/NVIDIA/nvidia-resiliency-ext.git@v0.5.0",
     "tqdm",
     "einops~=0.8",
     "tensorstore~=0.1,!=0.1.46,!=0.1.72",
@@ -60,7 +60,6 @@
     "mamba-ssm~=2.2",
     "causal-conv1d~=1.5",
     "flash-linear-attention~=0.4.0",
-    "nv-grouped-gemm~=1.1",
     "megatron-energon[av_decode]~=6.0",
     "av",
     "flashinfer-python~=0.5.0",
diff --git a/examples/configs/recipes/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.yaml b/examples/configs/recipes/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.yaml
index 071f90d077..00e47fac81 100644
--- a/examples/configs/recipes/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.yaml
+++ b/examples/configs/recipes/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.yaml
@@ -9,7 +9,7 @@ policy:
     name: ${policy.model_name}
   train_global_batch_size: 32
   train_micro_batch_size: 1
-  max_total_sequence_length: 6000
+  max_total_sequence_length: 3200
   dtensor_cfg:
     enabled: false
   megatron_cfg:
@@ -26,8 +26,6 @@ policy:
       lr: 1.0e-06
       min_lr: 1.0e-06
       adam_beta2: 0.999
-      use_distributed_optimizer: false
-      use_precision_aware_optimizer: false
     scheduler:
       lr_warmup_iters: 10
       lr_warmup_init: 1.0e-11
diff --git a/examples/configs/recipes/llm/grpo-dapomath17k-dsv3-megatron.yaml b/examples/configs/recipes/llm/grpo-dapomath17k-dsv3-megatron.yaml
index d1d16a9213..1cd7b4740f 100644
--- a/examples/configs/recipes/llm/grpo-dapomath17k-dsv3-megatron.yaml
+++ b/examples/configs/recipes/llm/grpo-dapomath17k-dsv3-megatron.yaml
@@ -30,6 +30,7 @@ policy:
     sequence_parallel: true
     moe_permute_fusion: true
     apply_rope_fusion: false
+    gradient_accumulation_fusion: false
     moe_enable_deepep: true
     moe_token_dispatcher_type: flex
     optimizer:
diff --git a/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron-fp8-e2e.yaml b/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron-fp8-e2e.yaml
index 54b8d6671f..d768a61bf6 100644
--- a/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron-fp8-e2e.yaml
+++ b/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron-fp8-e2e.yaml
@@ -26,6 +26,7 @@ policy:
     num_layers_in_first_pipeline_stage: 7
     num_layers_in_last_pipeline_stage: 6
     apply_rope_fusion: false
+    gradient_accumulation_fusion: false
     fp8_cfg:
       enabled: true
     optimizer:
diff --git a/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron.yaml b/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron.yaml
index 83ea6128ef..1e912de130 100644
--- a/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron.yaml
+++ b/examples/configs/recipes/llm/grpo-moonlight-16ba3b-4n8g-megatron.yaml
@@ -27,6 +27,7 @@ policy:
     num_layers_in_first_pipeline_stage: 7
     num_layers_in_last_pipeline_stage: 6
     apply_rope_fusion: false
+    gradient_accumulation_fusion: false
     optimizer:
       lr: 1.0e-06
     scheduler:
diff --git a/nemo_rl/models/dtensor/parallelize.py b/nemo_rl/models/dtensor/parallelize.py
index a4781a0afd..85d259b88d 100644
--- a/nemo_rl/models/dtensor/parallelize.py
+++ b/nemo_rl/models/dtensor/parallelize.py
@@ -489,9 +489,10 @@ def _parallelize_nm5_h(
         "mixer.down_proj": RowwiseParallel(),
     }
 
-    # Native transformers NemotronH uses model.model.layers, custom uses model.backbone.layers
-    inner_model = model.backbone if hasattr(model, "backbone") else model.model
+    # NemotronH uses .backbone (trust_remote_code) or .model (native transformers >= 5.3.0)
+    inner_model = getattr(model, "backbone", model.model)
     layers: torch.nn.ModuleList = inner_model.layers
+
     parallelize_module(model, tp_mesh, model_tp_plan)
 
     for layer in inner_model.layers:
@@ -522,7 +523,7 @@ def _parallelize_nm5_h(
 
     # do not reshard after forward for root model
     # because its parameters will be used in backward immediately
-    return fully_shard(
+    result = fully_shard(
         model,
         mesh=dp_mesh,
         mp_policy=mp_policy,
@@ -530,6 +531,14 @@ def _parallelize_nm5_h(
         reshard_after_forward=False,
     )
 
+    # Register .model so the native transformers forward() (self.model(...)) resolves
+    # correctly after FSDP2 wrapping, regardless of whether the class uses .backbone
+    # (trust_remote_code) or .model (native transformers). kernel_patches.py wraps
+    # the native forward which always calls self.model(...).
+    result.register_module("model", inner_model)
+
+    return result
+
 
 def _parallelize_model(
     model: Union[
diff --git a/nemo_rl/models/megatron/community_import.py b/nemo_rl/models/megatron/community_import.py
index 2e52a54f9c..334501dc53 100644
--- a/nemo_rl/models/megatron/community_import.py
+++ b/nemo_rl/models/megatron/community_import.py
@@ -75,6 +75,12 @@ def import_model_from_hf_name(
         ]
         model_provider.pipeline_dtype = megatron_config["pipeline_dtype"]
         model_provider.sequence_parallel = megatron_config["sequence_parallel"]
+        if (
+            gradient_accumulation_fusion := megatron_config.get(
+                "gradient_accumulation_fusion"
+            )
+        ) is not None:
+            model_provider.gradient_accumulation_fusion = gradient_accumulation_fusion
     model_provider.finalize()
 
     from megatron.core import parallel_state
diff --git a/pyproject.toml b/pyproject.toml
index 73b2db37a5..17830c391c 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -43,18 +43,18 @@ dependencies = [
   "sympy>=1.14.0",
   "pillow>=11.3.0",
   "torchvision==0.25.0",
-  "transformers==5.2.0",
+  "transformers==5.3.0",
   "num2words>=0.5.14",                                                                                                # for SmolVLM
   "mlflow>=3.5.0,<3.6.0",
   "nvidia-nvshmem-cu12; sys_platform == 'linux' and (platform_machine == 'x86_64' or platform_machine == 'aarch64')", # for deep_ep build
   "swanlab",
   "pyzmq",
   "decord2",
-  "nvidia-resiliency-ext",
-  "nccl4py",                                                                                                          # for non-colocated refit
-  "cuda-bindings",                                                                                                    # for non-colocated refit
-  "pybase64",                                                                                                         # for sglang refit
-  "nvidia-cudnn-cu12==9.19.0.56",                                                                                     # for transformer-engine no build isolation
+
+  "nccl4py",                      # for non-colocated refit
+  "cuda-bindings",                # for non-colocated refit
+  "pybase64",                     # for sglang refit
+  "nvidia-cudnn-cu12==9.19.0.56", # for transformer-engine no build isolation
 ]
 
 [project.optional-dependencies]
@@ -65,7 +65,7 @@ automodel = [
   # https://github.com/NVIDIA/TransformerEngine/blob/v2.3/transformer_engine/pytorch/attention/dot_product_attention/utils.py#L108
   # https://github.com/facebookresearch/xformers/blob/8354497deb2c04c67fbb2e2ad911e86530da0e90/xformers/ops/fmha/flash.py#L76
   "flash-attn==2.8.1",
-  "transformers>=5.0.0",
+  "transformers>=5.3.0",
   "mamba-ssm",
   "causal-conv1d",
   "nv-grouped-gemm",
@@ -258,7 +258,7 @@ override-dependencies = [
   "opentelemetry-api>=1.33.1",
   # vLLM 0.17.0 code is compatible with transformers v5 but the PyPI metadata still declares <5.
   # Override until vllm officially relaxes the constraint (https://github.com/vllm-project/vllm/issues/30466).
-  "transformers==5.2.0",
+  "transformers==5.3.0",
 ]
 # CVE fixes
 constraint-dependencies = [
@@ -356,6 +356,7 @@ name = "nv-grouped-gemm"
 version = "v1.1.4.post7"
 requires-dist = ["setuptools", "wheel", "torch", "numpy"]
 
+
 [[tool.uv.dependency-metadata]]
 name = "sgl-kernel"
 # This version has to match the version in the commit/rev/tag used
@@ -423,7 +424,7 @@ requires-dist = [
   "torchcodec==0.8.0 ; sys_platform != 'linux' or (sys_platform == 'linux' and platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l')",
   "torchvision",
   "tqdm",
-  "transformers==5.2.0",
+  "transformers==5.3.0",
   "uvicorn",
   "uvloop",
   "xgrammar==0.1.27",
@@ -440,9 +441,15 @@ name = "megatron-bridge"
 # Must stay in sync with 3rdparty/Megatron-Bridge-workspace/setup.py::CACHED_DEPENDENCIES.
 version = "0.0.0"
 requires-dist = [
-  "transformers>=5.0.0,<=5.2.0",
+  "transformers>=5.0.0,<=5.3.0",
+  "peft>=0.18.1",
   "datasets>=2.20.0",
   "accelerate",
+  "diffusers>=0.36.0",
+  "peft>=0.18.0",
+  "einops",
+  "imageio",
+  "imageio-ffmpeg",
   "omegaconf>=2.3.0",
   "tensorboard>=2.19.0",
   "typing-extensions",
@@ -459,12 +466,13 @@ requires-dist = [
   # TODO(https://github.com/NVIDIA-NeMo/RL/issues/2111): upgrade to core_cu13 when we move to CUDA 13
   "transformer-engine[pytorch,core_cu12]",
   "mamba-ssm",
-  "nvidia-resiliency-ext~=0.4.1",
+  "nvidia-resiliency-ext~=0.5.0",
   "causal-conv1d",
   "flash-linear-attention",
   "timm",
   "open-clip-torch>=3.2.0",
   "mlflow>=3.5.0",
+  "comet-ml>=3.50.0",
   "torch>=2.6.0",
 ]
 
diff --git a/tests/test_suites/llm/dpo-nanov3-30B3AB-1n8g-fsdp8ep8-automodel.sh b/tests/test_suites/llm/dpo-nanov3-30B3AB-1n8g-fsdp8ep8-automodel.sh
index 988fc14f37..faefa7bf3f 100755
--- a/tests/test_suites/llm/dpo-nanov3-30B3AB-1n8g-fsdp8ep8-automodel.sh
+++ b/tests/test_suites/llm/dpo-nanov3-30B3AB-1n8g-fsdp8ep8-automodel.sh
@@ -35,10 +35,10 @@ uv run tests/json_dump_tb_logs.py $LOG_DIR --output_path $JSON_METRICS
 if [[ $(jq 'to_entries | .[] | select(.key == "train/loss") | .value | keys | map(tonumber) | max' $JSON_METRICS) -ge $MAX_STEPS ]]; then
     uv run tests/check_metrics.py $JSON_METRICS \
         'data["train/loss"]["1"] < 0.69316' \
-        'data["train/loss"]["15"] < 0.63263' \
+        'data["train/loss"]["11"] < 0.53' \
         'data["train/preference_loss"]["1"] > 0.69314' \
         'data["train/preference_loss"]["1"] < 0.69316' \
-        'data["train/preference_loss"]["15"] < 0.63263' \
+        'data["train/preference_loss"]["11"] < 0.53' \
         'mean(data["timing/train/total_step_time"], -5, -1) < 5'
 
     # Clean up checkpoint directory after successful run to save space.
diff --git a/tests/test_suites/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh b/tests/test_suites/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
index dbab6cb2f7..ec44a8a960 100755
--- a/tests/test_suites/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
+++ b/tests/test_suites/llm/dpo-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
@@ -35,8 +35,9 @@ uv run tests/json_dump_tb_logs.py $LOG_DIR --output_path $JSON_METRICS
 if [[ $(jq 'to_entries | .[] | select(.key == "train/loss") | .value | keys | map(tonumber) | max' $JSON_METRICS) -ge $MAX_STEPS ]]; then
     # Smoke checks: run completed and loss is finite/reasonable.
     uv run tests/check_metrics.py $JSON_METRICS \
-        'data["train/loss"]["10"] > 0.0' \
-        'data["train/loss"]["10"] < 20.0'
+        'data["train/loss"]["1"] < 0.7' \
+        'data["train/loss"]["10"] < 0.7' \
+        'data["train/accuracy"]["10"] > 0.56'
 
     # Clean up checkpoint directory after successful run to save space.
     rm -rf "$CKPT_DIR"
diff --git a/tests/test_suites/llm/sft-llama3.1-8b-1n8g-fsdp2tp4-dynamicbatch.sh b/tests/test_suites/llm/sft-llama3.1-8b-1n8g-fsdp2tp4-dynamicbatch.sh
index 65362c2eb1..1e31b56674 100755
--- a/tests/test_suites/llm/sft-llama3.1-8b-1n8g-fsdp2tp4-dynamicbatch.sh
+++ b/tests/test_suites/llm/sft-llama3.1-8b-1n8g-fsdp2tp4-dynamicbatch.sh
@@ -39,7 +39,7 @@ if [[ $(jq 'to_entries | .[] | select(.key == "train/loss") | .value | keys | ma
 	    'data["train/loss"]["1"] < 0.6' \
         'data["train/loss"]["250"] < 0.36' \
         'max(data["ray/node.0.gpu.0.mem_gb"]) < 75' \
-        'mean(data["timing/train/total_step_time"], -6, -1) < 10'
+        'mean(data["timing/train/total_step_time"], -6, -1) < 10.5'
 
     # Clean up checkpoint directory after successful run to save space.
     rm -rf "$CKPT_DIR"
diff --git a/tests/test_suites/llm/sft-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh b/tests/test_suites/llm/sft-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
index 261a0e851c..ed26ecb2c2 100755
--- a/tests/test_suites/llm/sft-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
+++ b/tests/test_suites/llm/sft-qwen2.5-math7b-1n8g-megatron_chunked_linear_ce_loss.sh
@@ -36,8 +36,8 @@ uv run tests/json_dump_tb_logs.py $LOG_DIR --output_path $JSON_METRICS
 if [[ $(jq 'to_entries | .[] | select(.key == "train/loss") | .value | keys | map(tonumber) | max' $JSON_METRICS) -ge $MAX_STEPS ]]; then
     # Smoke checks: run completed and loss is finite/reasonable.
     uv run tests/check_metrics.py $JSON_METRICS \
-        'data["train/loss"]["10"] > 0.0' \
-        'data["train/loss"]["10"] < 20.0'
+        'data["train/loss"]["1"] < 0.49' \
+        'data["train/loss"]["10"] < 0.45'
 
     # Clean up checkpoint directory after successful run to save space.
     rm -rf "$CKPT_DIR"
diff --git a/tests/unit/prepare_unit_test_assets.py b/tests/unit/prepare_unit_test_assets.py
index 488562b0b8..0391d117f3 100644
--- a/tests/unit/prepare_unit_test_assets.py
+++ b/tests/unit/prepare_unit_test_assets.py
@@ -36,33 +36,27 @@
 def build_tiny_nemotron5_h_checkpoint(model_path: str) -> None:
     import shutil
 
-    import transformers
-    from transformers import AutoConfig, AutoModelForCausalLM, AutoTokenizer
-
-    assert transformers.__version__ < "5.3.0", (
-        "NemotronHConfig is supported in transformers 5.3.0 or later, use NemotronHConfig instead"
+    from transformers import AutoModelForCausalLM, AutoTokenizer
+    from transformers.models.nemotron_h import NemotronHConfig
+
+    config = NemotronHConfig(
+        layers_block_type=["mamba", "attention", "mamba"],
+        num_hidden_layers=3,
+        intermediate_size=32,
+        hidden_size=256,
+        num_attention_heads=8,
+        mamba_num_heads=8,
+        num_key_value_heads=8,
+        n_groups=1,
+        vocab_size=131072,
     )
 
-    config = AutoConfig.from_pretrained(
-        "nvidia/Nemotron-H-8B-Base-8K", trust_remote_code=True
-    )
-    config.hybrid_override_pattern = "M*-"
-    config.num_hidden_layers = 3
-    config.intermediate_size = 32
-    config.hidden_size = 256
-    config.num_attention_heads = 8
-    config.mamba_num_heads = 8
-    config.num_key_value_heads = 8
-    config.n_groups = 1
-
-    model = AutoModelForCausalLM.from_config(config, trust_remote_code=True)
+    model = AutoModelForCausalLM.from_config(config)
+
     tokenizer = AutoTokenizer.from_pretrained(
         "nvidia/Nemotron-H-8B-Base-8K", trust_remote_code=True
     )
 
-    # Disable tied weights for transformers 5.2.0 to avoid error
-    model._tied_weights_keys = None
-
     shutil.rmtree(model_path, ignore_errors=True)
     model.save_pretrained(model_path)
     tokenizer.save_pretrained(model_path)
diff --git a/uv.lock b/uv.lock
index e72d2c4c72..27ddd02b2a 100644
--- a/uv.lock
+++ b/uv.lock
@@ -116,7 +116,7 @@ overrides = [
     { name = "torch", marker = "sys_platform == 'darwin'", specifier = "==2.10.0", index = "https://pypi.org/simple" },
     { name = "torchaudio", specifier = "==2.10.0" },
     { name = "transformer-engine", extras = ["pytorch"], specifier = "==2.12.0" },
-    { name = "transformers", specifier = "==5.2.0" },
+    { name = "transformers", specifier = "==5.3.0" },
 ]
 
 [[manifest.dependency-metadata]]
@@ -146,7 +146,7 @@ requires-dist = ["torch", "packaging", "ninja", "causal-conv1d"]
 [[manifest.dependency-metadata]]
 name = "megatron-bridge"
 version = "0.0.0"
-requires-dist = ["transformers>=5.0.0,<=5.2.0", "datasets>=2.20.0", "accelerate", "omegaconf>=2.3.0", "tensorboard>=2.19.0", "typing-extensions", "rich", "wandb>=0.25.0", "six>=1.17.0", "regex>=2024.11.6", "pyyaml>=6.0.2", "tqdm>=4.67.1", "hydra-core>1.3,<=1.3.2", "qwen-vl-utils", "transformer-engine[pytorch,core-cu12]", "mamba-ssm", "nvidia-resiliency-ext~=0.4.1", "causal-conv1d", "flash-linear-attention", "timm", "open-clip-torch>=3.2.0", "mlflow>=3.5.0", "torch>=2.6.0"]
+requires-dist = ["transformers>=5.0.0,<=5.3.0", "peft>=0.18.1", "datasets>=2.20.0", "accelerate", "diffusers>=0.36.0", "peft>=0.18.0", "einops", "imageio", "imageio-ffmpeg", "omegaconf>=2.3.0", "tensorboard>=2.19.0", "typing-extensions", "rich", "wandb>=0.25.0", "six>=1.17.0", "regex>=2024.11.6", "pyyaml>=6.0.2", "tqdm>=4.67.1", "hydra-core>1.3,<=1.3.2", "qwen-vl-utils", "transformer-engine[pytorch,core-cu12]", "mamba-ssm", "nvidia-resiliency-ext~=0.5.0", "causal-conv1d", "flash-linear-attention", "timm", "open-clip-torch>=3.2.0", "mlflow>=3.5.0", "comet-ml>=3.50.0", "torch>=2.6.0"]
 
 [[manifest.dependency-metadata]]
 name = "nv-grouped-gemm"
@@ -161,7 +161,7 @@ requires-dist = ["torch", "scikit-build-core", "wheel"]
 [[manifest.dependency-metadata]]
 name = "sglang"
 version = "0.5.7.dev0"
-requires-dist = ["ipython", "aiohttp", "apache-tvm-ffi>=0.1.5,<0.2", "anthropic>=0.20.0", "blobfile==3.0.0", "build", "compressed-tensors", "cuda-python==12.9", "decord2", "datasets", "einops", "fastapi", "flashinfer-python==0.6.4", "flashinfer-cubin==0.6.4", "gguf", "interegular", "llguidance>=0.7.11,<0.8.0", "modelscope", "msgspec", "ninja", "numpy", "nvidia-cutlass-dsl>=4.3.4", "nvidia-ml-py", "openai-harmony==0.0.4", "openai==2.6.1", "orjson", "outlines==0.1.11", "packaging", "partial-json-parser", "pillow", "prometheus-client>=0.20.0", "psutil", "py-spy", "pybase64", "pydantic", "python-multipart", "pyzmq>=25.1.2", "quack-kernels==0.2.4", "requests", "scipy", "sentencepiece", "setproctitle", "sgl-kernel==0.3.21", "soundfile==0.13.1", "tiktoken", "timm==1.0.16", "torch-memory-saver==0.0.9", "torch==2.9.1", "torchao==0.9.0", "torchaudio==2.9.1", "torchcodec==0.8.0 ; (platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l') or sys_platform != 'linux'", "torchvision", "tqdm", "transformers==5.2.0", "uvicorn", "uvloop", "xgrammar==0.1.27", "smg-grpc-proto>=0.3.3", "grpcio>=1.78.0", "grpcio-reflection>=1.78.0", "grpcio-health-checking>=1.78.0"]
+requires-dist = ["ipython", "aiohttp", "apache-tvm-ffi>=0.1.5,<0.2", "anthropic>=0.20.0", "blobfile==3.0.0", "build", "compressed-tensors", "cuda-python==12.9", "decord2", "datasets", "einops", "fastapi", "flashinfer-python==0.6.4", "flashinfer-cubin==0.6.4", "gguf", "interegular", "llguidance>=0.7.11,<0.8.0", "modelscope", "msgspec", "ninja", "numpy", "nvidia-cutlass-dsl>=4.3.4", "nvidia-ml-py", "openai-harmony==0.0.4", "openai==2.6.1", "orjson", "outlines==0.1.11", "packaging", "partial-json-parser", "pillow", "prometheus-client>=0.20.0", "psutil", "py-spy", "pybase64", "pydantic", "python-multipart", "pyzmq>=25.1.2", "quack-kernels==0.2.4", "requests", "scipy", "sentencepiece", "setproctitle", "sgl-kernel==0.3.21", "soundfile==0.13.1", "tiktoken", "timm==1.0.16", "torch-memory-saver==0.0.9", "torch==2.9.1", "torchao==0.9.0", "torchaudio==2.9.1", "torchcodec==0.8.0 ; (platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l') or sys_platform != 'linux'", "torchvision", "tqdm", "transformers==5.3.0", "uvicorn", "uvloop", "xgrammar==0.1.27", "smg-grpc-proto>=0.3.3", "grpcio>=1.78.0", "grpcio-reflection>=1.78.0", "grpcio-health-checking>=1.78.0"]
 
 [[package]]
 name = "absl-py"
@@ -1094,6 +1094,32 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/e2/98/0d791b3d1eaed89d7d370b5cf9b8079b124da0545559417f394ba21b5532/colorful-0.5.7-py2.py3-none-any.whl", hash = "sha256:495dd3a23151a9568cee8a90fc1174c902ad7ef06655f50b6bddf9e80008da69", size = 201475, upload-time = "2025-06-30T15:24:02.693Z" },
 ]
 
+[[package]]
+name = "comet-ml"
+version = "3.57.3"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "dulwich" },
+    { name = "everett", extra = ["ini"], marker = "extra == 'extra-7-nemo-rl-mcore' or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-automodel' and extra != 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra != 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm')" },
+    { name = "jsonschema" },
+    { name = "psutil" },
+    { name = "python-box" },
+    { name = "requests" },
+    { name = "requests-toolbelt" },
+    { name = "rich" },
+    { name = "semantic-version" },
+    { name = "sentry-sdk" },
+    { name = "setuptools" },
+    { name = "simplejson" },
+    { name = "urllib3" },
+    { name = "wrapt" },
+    { name = "wurlitzer" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b8/e3/39e6a2ecb0f83bacc414c23cbab86a330540396607fc6894588a1d71e092/comet_ml-3.57.3.tar.gz", hash = "sha256:84be2337366d598e8a299eb56e2efcffb859d7de7957148f81f124d3ef02728d", size = 585296, upload-time = "2026-03-12T14:23:22.371Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/6a/17/c875fe4d53286f8664d3edf7185b488cddfca6c208ba978aa34550e9d10d/comet_ml-3.57.3-py3-none-any.whl", hash = "sha256:4eedf5da98a000466759adbd7e96d6f44004ce1bd844e9bc03d6ec5f86583249", size = 786246, upload-time = "2026-03-12T14:23:20.942Z" },
+]
+
 [[package]]
 name = "compressed-tensors"
 version = "0.13.0"
@@ -1110,6 +1136,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/0b/b5/61ac2563c62490922b603c09113a083fd74af3630ec3931e769484d6dcb5/compressed_tensors-0.13.0-py3-none-any.whl", hash = "sha256:3518799c9baf034eb642efb551db6b0537b8713d45a64fe4def26f7f8d6cabec", size = 192620, upload-time = "2025-12-16T16:03:53.041Z" },
 ]
 
+[[package]]
+name = "configobj"
+version = "5.0.9"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f5/c4/c7f9e41bc2e5f8eeae4a08a01c91b2aea3dfab40a3e14b25e87e7db8d501/configobj-5.0.9.tar.gz", hash = "sha256:03c881bbf23aa07bccf1b837005975993c4ab4427ba57f959afdd9d1a2386848", size = 101518, upload-time = "2024-09-21T12:47:46.315Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a6/c4/0679472c60052c27efa612b4cd3ddd2a23e885dcdc73461781d2c802d39e/configobj-5.0.9-py2.py3-none-any.whl", hash = "sha256:1ba10c5b6ee16229c79a05047aeda2b55eb4e80d7c7d8ecf17ec1ca600c79882", size = 35615, upload-time = "2024-11-26T14:03:32.972Z" },
+]
+
 [[package]]
 name = "contourpy"
 version = "1.3.3"
@@ -1671,6 +1706,26 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/d1/ae/afb1487556e2dc827a17097aac8158a25b433a345386f0e249f6d2694ccb/devtools-0.12.2-py3-none-any.whl", hash = "sha256:c366e3de1df4cdd635f1ad8cbcd3af01a384d7abda71900e68d43b04eb6aaca7", size = 19411, upload-time = "2023-09-03T16:56:59.049Z" },
 ]
 
+[[package]]
+name = "diffusers"
+version = "0.37.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "filelock" },
+    { name = "httpx" },
+    { name = "huggingface-hub" },
+    { name = "importlib-metadata" },
+    { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" } },
+    { name = "pillow" },
+    { name = "regex" },
+    { name = "requests" },
+    { name = "safetensors" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/58/3b/01d0ff800b811c5ad8bba682f4c6abf1d7071cd81464c01724333fefb7ba/diffusers-0.37.0.tar.gz", hash = "sha256:408789af73898585f525afd07ca72b3955affea4216a669558e9f59b5b1fe704", size = 4141136, upload-time = "2026-03-05T14:58:39.704Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f7/55/586a3a2b9c95f371c9c3cb048c3cac15aedcce8d6d53ebd6bbc46860722d/diffusers-0.37.0-py3-none-any.whl", hash = "sha256:7eab74bf896974250b5e1027cae813aba1004f02d97c9b44891b83713386aa08", size = 5000449, upload-time = "2026-03-05T14:58:37.361Z" },
+]
+
 [[package]]
 name = "dill"
 version = "0.3.8"
@@ -1754,6 +1809,32 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/8f/d7/9322c609343d929e75e7e5e6255e614fcc67572cfd083959cdef3b7aad79/docutils-0.21.2-py3-none-any.whl", hash = "sha256:dafca5b9e384f0e419294eb4d2ff9fa826435bf15f15b7bd45723e8ad76811b2", size = 587408, upload-time = "2024-04-23T18:57:14.835Z" },
 ]
 
+[[package]]
+name = "dulwich"
+version = "0.25.2"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "urllib3" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/ba/b8/9c9bc6ac66007f8090b1da9079c0e4bbea5aa9583c3c12098e0f11462dd5/dulwich-0.25.2.tar.gz", hash = "sha256:bca22c8aa4cbecbe8493b76e3fd6101513f09cf405cd9b92e116a48d9469e55a", size = 1126499, upload-time = "2026-01-11T22:04:47.667Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/da/8a/4ec87df697cf1af9172b015e1256ca93856d9454d7e24a4f9168d3667892/dulwich-0.25.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ce00c68c4fcd7ea53641153a69aab9a010ae140387a39f13e9ecf05f60fefd77", size = 1318435, upload-time = "2026-01-11T22:04:21.97Z" },
+    { url = "https://files.pythonhosted.org/packages/e1/fe/1260a7217eb439bae33bae3af98b84ed53e0601e19bd87e580df09650021/dulwich-0.25.2-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:6ece907b40f503c68e27bd77c71d3de25ac5c6256c43b82f7843232e7769cebd", size = 1395034, upload-time = "2026-01-11T22:04:23.384Z" },
+    { url = "https://files.pythonhosted.org/packages/3f/24/e8cec93df1bfba4087919842a0754b50f0c6e605d620976d5d8625229caa/dulwich-0.25.2-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:e2d5cc06cc25d88f87fd966bee74c62903473f81a1646323bf1e4fe8fec4b797", size = 1423110, upload-time = "2026-01-11T22:04:24.937Z" },
+    { url = "https://files.pythonhosted.org/packages/4e/4b/f4ef7c2dcf7b47c27518461e0acf32eaf76fd357a1aa02ce3de0f1b04578/dulwich-0.25.2-cp312-cp312-win32.whl", hash = "sha256:62c7fe4931a5457745aaa263dea6388a6334ba03e65990fadd10b1857f5ad741", size = 982792, upload-time = "2026-01-11T22:04:26.929Z" },
+    { url = "https://files.pythonhosted.org/packages/87/2b/bee92d4c4dc8ccfdbe64a87464e5970c78ea9b201c7d57f15342330d32de/dulwich-0.25.2-cp312-cp312-win_amd64.whl", hash = "sha256:3977d089e4c68fc1589457d7a19a7637a1d8f173702f18eb1c198bb4d34e52b0", size = 1000183, upload-time = "2026-01-11T22:04:29.013Z" },
+    { url = "https://files.pythonhosted.org/packages/82/6b/a2f422be19ddbbd6a56477e0a40a8ea7c58628467e655143c249d8c320cf/dulwich-0.25.2-cp313-cp313-android_21_arm64_v8a.whl", hash = "sha256:46bfb777b33f2906c9800ce8c8ad0ea0530c1c2d1145eab6d42c40de29f73efa", size = 1419859, upload-time = "2026-01-11T22:04:30.721Z" },
+    { url = "https://files.pythonhosted.org/packages/2e/ee/d0954d64322955d8cd1c482263925ca75378e640851218cb14ffe16aae07/dulwich-0.25.2-cp313-cp313-android_21_x86_64.whl", hash = "sha256:2a845afcd30d049a222240f9efdec6b95c2b6fd839564777061e6209e54c3ffc", size = 1419852, upload-time = "2026-01-11T22:04:32.669Z" },
+    { url = "https://files.pythonhosted.org/packages/4e/cf/07f6a26837e79b5f6483fdc77f79f661aa59ed86fcc13e61bc233d95e6d4/dulwich-0.25.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:26bfe8c35680dd0cf71ce724e0f00401a439a332e8bd90a82e556ab2cb3a68e6", size = 1318305, upload-time = "2026-01-11T22:04:34.142Z" },
+    { url = "https://files.pythonhosted.org/packages/3d/2a/aa784b51554d005a35ff78859424e9b69e9c4124533e5063ebe4161ad10c/dulwich-0.25.2-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:e7ec5bc1e769b19312d1ae431981096aa046925e9cb278b8efff6bebdb679b12", size = 1394619, upload-time = "2026-01-11T22:04:35.832Z" },
+    { url = "https://files.pythonhosted.org/packages/89/93/4e95a9a92fbc01f5d1bf996b6393c3dabde26031c1c8100355c189fec8f4/dulwich-0.25.2-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:ab15cc01c19bb1b258f6843470637bc5f2d886b8244bb48f8da8ee3d766bcf10", size = 1422512, upload-time = "2026-01-11T22:04:37.481Z" },
+    { url = "https://files.pythonhosted.org/packages/c4/7e/d7b1b0c83457e2ad75cee64e1390151ac25ac89597e5a8f6530137e1c1fd/dulwich-0.25.2-cp313-cp313-win32.whl", hash = "sha256:a7ccd96e3beb93df7458191f0aadad6e76ab78f09452f867fc06cd4f99423c7e", size = 983597, upload-time = "2026-01-11T22:04:39.064Z" },
+    { url = "https://files.pythonhosted.org/packages/1a/4a/3cb5178b49a8be5d311276af33a8e6f8d3cce0f6410b6c03ab99b96e74eb/dulwich-0.25.2-cp313-cp313-win_amd64.whl", hash = "sha256:2f84e6501702877ecc1c1a8710c745942d86d2f55cbfeaf99377100e4c16139a", size = 1000141, upload-time = "2026-01-11T22:04:40.604Z" },
+    { url = "https://files.pythonhosted.org/packages/82/ec/494f14d73346309e2e03fdd1fa82618d91bbc59423bbe8a6f6a7b20186ee/dulwich-0.25.2-cp314-cp314-android_24_arm64_v8a.whl", hash = "sha256:b1b54442dd8171fc5a1e0d5efc7d72b8192c88f738ee9d72e7aa82bf9d630832", size = 1437740, upload-time = "2026-01-11T22:04:42.297Z" },
+    { url = "https://files.pythonhosted.org/packages/c8/48/8448a48054f61e1c4c7c42f2ab29cdb576451545d2843651f69802ff15fb/dulwich-0.25.2-cp314-cp314-android_24_x86_64.whl", hash = "sha256:0ac0b70a970fac9b9c161ce2f1472915656c91e8fdb2dcfb1b5f84e6a127a184", size = 1437733, upload-time = "2026-01-11T22:04:43.978Z" },
+    { url = "https://files.pythonhosted.org/packages/87/eb/153b2b32dca090e956a1e512293db3c7c144db50da439373d1be56880512/dulwich-0.25.2-py3-none-any.whl", hash = "sha256:19dd5a0e08a47483be7f404e2555136a9ebaf70781fee3280457f8e2d65b2388", size = 650045, upload-time = "2026-01-11T22:04:45.398Z" },
+]
+
 [[package]]
 name = "ebmlite"
 version = "3.4.1"
@@ -1796,6 +1877,20 @@ dependencies = [
     { name = "typing-extensions" },
 ]
 
+[[package]]
+name = "everett"
+version = "3.1.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/c0/b4/c7c61c0b243c4277d19299cd1bccee8b2b57d04073c0d8625799fe47f5c9/everett-3.1.0.tar.gz", hash = "sha256:46175da5bcb06c193aa129e59714bca981344ff067c3a8bc2e625bc0b3dc01f6", size = 73796, upload-time = "2022-10-26T15:15:00.651Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/91/9a/d882fd7562208456236fb2e62b762bf16fbc9ecde842bb871f676ca0f7e1/everett-3.1.0-py2.py3-none-any.whl", hash = "sha256:db13891b849e45e54faea93ee79881d12458c5378f5b9b7f806eeff03ce1de3c", size = 35702, upload-time = "2022-10-26T15:14:58.698Z" },
+]
+
+[package.optional-dependencies]
+ini = [
+    { name = "configobj" },
+]
+
 [[package]]
 name = "execnet"
 version = "2.1.2"
@@ -2759,6 +2854,33 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/23/28/96711503245339084c8086b892c47415895eba49782d6cc52d9f4ee50301/ijson-3.5.0-cp314-cp314t-win_amd64.whl", hash = "sha256:4f24b78d4ef028d17eb57ad1b16c0aed4a17bdd9badbf232dc5d9305b7e13854", size = 58965, upload-time = "2026-02-24T03:58:11.278Z" },
 ]
 
+[[package]]
+name = "imageio"
+version = "2.37.3"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" } },
+    { name = "pillow" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b1/84/93bcd1300216ea50811cee96873b84a1bebf8d0489ffaf7f2a3756bab866/imageio-2.37.3.tar.gz", hash = "sha256:bbb37efbfc4c400fcd534b367b91fcd66d5da639aaa138034431a1c5e0a41451", size = 389673, upload-time = "2026-03-09T11:31:12.573Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/49/fa/391e437a34e55095173dca5f24070d89cbc233ff85bf1c29c93248c6588d/imageio-2.37.3-py3-none-any.whl", hash = "sha256:46f5bb8522cd421c0f5ae104d8268f569d856b29eb1a13b92829d1970f32c9f0", size = 317646, upload-time = "2026-03-09T11:31:10.771Z" },
+]
+
+[[package]]
+name = "imageio-ffmpeg"
+version = "0.6.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/44/bd/c3343c721f2a1b0c9fc71c1aebf1966a3b7f08c2eea8ed5437a2865611d6/imageio_ffmpeg-0.6.0.tar.gz", hash = "sha256:e2556bed8e005564a9f925bb7afa4002d82770d6b08825078b7697ab88ba1755", size = 25210, upload-time = "2025-01-16T21:34:32.747Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/da/58/87ef68ac83f4c7690961bce288fd8e382bc5f1513860fc7f90a9c1c1c6bf/imageio_ffmpeg-0.6.0-py3-none-macosx_10_9_intel.macosx_10_9_x86_64.whl", hash = "sha256:9d2baaf867088508d4a3458e61eeb30e945c4ad8016025545f66c4b5aaef0a61", size = 24932969, upload-time = "2025-01-16T21:34:20.464Z" },
+    { url = "https://files.pythonhosted.org/packages/40/5c/f3d8a657d362cc93b81aab8feda487317da5b5d31c0e1fdfd5e986e55d17/imageio_ffmpeg-0.6.0-py3-none-macosx_11_0_arm64.whl", hash = "sha256:b1ae3173414b5fc5f538a726c4e48ea97edc0d2cdc11f103afee655c463fa742", size = 21113891, upload-time = "2025-01-16T21:34:00.277Z" },
+    { url = "https://files.pythonhosted.org/packages/33/e7/1925bfbc563c39c1d2e82501d8372734a5c725e53ac3b31b4c2d081e895b/imageio_ffmpeg-0.6.0-py3-none-manylinux2014_aarch64.whl", hash = "sha256:1d47bebd83d2c5fc770720d211855f208af8a596c82d17730aa51e815cdee6dc", size = 25632706, upload-time = "2025-01-16T21:33:53.475Z" },
+    { url = "https://files.pythonhosted.org/packages/a0/2d/43c8522a2038e9d0e7dbdf3a61195ecc31ca576fb1527a528c877e87d973/imageio_ffmpeg-0.6.0-py3-none-manylinux2014_x86_64.whl", hash = "sha256:c7e46fcec401dd990405049d2e2f475e2b397779df2519b544b8aab515195282", size = 29498237, upload-time = "2025-01-16T21:34:13.726Z" },
+    { url = "https://files.pythonhosted.org/packages/a0/13/59da54728351883c3c1d9fca1710ab8eee82c7beba585df8f25ca925f08f/imageio_ffmpeg-0.6.0-py3-none-win32.whl", hash = "sha256:196faa79366b4a82f95c0f4053191d2013f4714a715780f0ad2a68ff37483cc2", size = 19652251, upload-time = "2025-01-16T21:34:06.812Z" },
+    { url = "https://files.pythonhosted.org/packages/2c/c6/fa760e12a2483469e2bf5058c5faff664acf66cadb4df2ad6205b016a73d/imageio_ffmpeg-0.6.0-py3-none-win_amd64.whl", hash = "sha256:02fa47c83703c37df6bfe4896aab339013f62bf02c5ebf2dce6da56af04ffc0a", size = 31246824, upload-time = "2025-01-16T21:34:28.6Z" },
+]
+
 [[package]]
 name = "imagesize"
 version = "1.4.1"
@@ -3444,14 +3566,20 @@ source = { editable = "3rdparty/Megatron-Bridge-workspace" }
 dependencies = [
     { name = "accelerate" },
     { name = "causal-conv1d" },
+    { name = "comet-ml" },
     { name = "datasets" },
+    { name = "diffusers" },
+    { name = "einops" },
     { name = "flash-linear-attention" },
     { name = "hydra-core" },
+    { name = "imageio" },
+    { name = "imageio-ffmpeg" },
     { name = "mamba-ssm" },
     { name = "mlflow" },
     { name = "nvidia-resiliency-ext" },
     { name = "omegaconf" },
     { name = "open-clip-torch" },
+    { name = "peft" },
     { name = "pyyaml" },
     { name = "qwen-vl-utils" },
     { name = "regex" },
@@ -3472,14 +3600,21 @@ dependencies = [
 requires-dist = [
     { name = "accelerate" },
     { name = "causal-conv1d" },
+    { name = "comet-ml", specifier = ">=3.50.0" },
     { name = "datasets", specifier = ">=2.20.0" },
+    { name = "diffusers", specifier = ">=0.36.0" },
+    { name = "einops" },
     { name = "flash-linear-attention" },
     { name = "hydra-core", specifier = ">1.3,<=1.3.2" },
+    { name = "imageio" },
+    { name = "imageio-ffmpeg" },
     { name = "mamba-ssm" },
     { name = "mlflow", specifier = ">=3.5.0" },
-    { name = "nvidia-resiliency-ext", specifier = "~=0.4.1" },
+    { name = "nvidia-resiliency-ext", specifier = "~=0.5.0" },
     { name = "omegaconf", specifier = ">=2.3.0" },
     { name = "open-clip-torch", specifier = ">=3.2.0" },
+    { name = "peft", specifier = ">=0.18.0" },
+    { name = "peft", specifier = ">=0.18.1" },
     { name = "pyyaml", specifier = ">=6.0.2" },
     { name = "qwen-vl-utils" },
     { name = "regex", specifier = ">=2024.11.6" },
@@ -3490,7 +3625,7 @@ requires-dist = [
     { name = "torch", specifier = ">=2.6.0" },
     { name = "tqdm", specifier = ">=4.67.1" },
     { name = "transformer-engine", extras = ["pytorch", "core-cu12"] },
-    { name = "transformers", specifier = ">=5.0.0,<=5.2.0" },
+    { name = "transformers", specifier = ">=5.0.0,<=5.3.0" },
     { name = "typing-extensions" },
     { name = "wandb", specifier = ">=0.25.0" },
 ]
@@ -3513,7 +3648,6 @@ dependencies = [
     { name = "multi-storage-client" },
     { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "(extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra != 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra != 'extra-7-nemo-rl-fsdp' and extra != 'extra-7-nemo-rl-mcore' and extra != 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
     { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-nemo-rl-automodel' or extra == 'extra-7-nemo-rl-mcore' or extra == 'extra-7-nemo-rl-sglang' or extra != 'extra-7-nemo-rl-vllm'" },
-    { name = "nv-grouped-gemm" },
     { name = "nvidia-modelopt" },
     { name = "nvidia-resiliency-ext" },
     { name = "nvtx" },
@@ -3546,9 +3680,8 @@ requires-dist = [
     { name = "megatron-energon", extras = ["av-decode"], specifier = "~=6.0" },
     { name = "multi-storage-client", specifier = "~=0.27" },
     { name = "numpy" },
-    { name = "nv-grouped-gemm", git = "https://github.com/fanshiqing/grouped_gemm?tag=v1.1.4.post7" },
     { name = "nvidia-modelopt", extras = ["torch"], marker = "sys_platform != 'darwin'" },
-    { name = "nvidia-resiliency-ext" },
+    { name = "nvidia-resiliency-ext", git = "https://github.com/NVIDIA/nvidia-resiliency-ext.git?rev=v0.5.0" },
     { name = "nvtx", specifier = "~=0.2" },
     { name = "onnxscript" },
     { name = "openai", extras = ["aiohttp"] },
@@ -3560,7 +3693,7 @@ requires-dist = [
     { name = "torch", marker = "sys_platform != 'darwin'", specifier = ">=2.6.0", index = "https://download.pytorch.org/whl/cu129" },
     { name = "torch", marker = "sys_platform == 'darwin'", specifier = ">=2.6.0", index = "https://pypi.org/simple" },
     { name = "tqdm" },
-    { name = "transformer-engine", extras = ["core-cu12", "pytorch"], specifier = ">=2.9.0a0,<2.12.0" },
+    { name = "transformer-engine", extras = ["core-cu12", "pytorch"] },
     { name = "wget" },
 ]
 
@@ -4177,7 +4310,7 @@ requires-dist = [
     { name = "torchvision", marker = "sys_platform != 'darwin' and sys_platform != 'linux' and extra == 'diffusion'", index = "https://download.pytorch.org/whl/cpu" },
     { name = "torchvision", marker = "sys_platform == 'linux' and extra == 'diffusion'", index = "https://download.pytorch.org/whl/cu129" },
     { name = "transformer-engine", extras = ["pytorch"], marker = "extra == 'cuda'", specifier = "<=2.11.0" },
-    { name = "transformers", specifier = ">=5.3.0" },
+    { name = "transformers", specifier = ">=5.3.0,<5.4.0" },
     { name = "wandb" },
 ]
 provides-extras = ["diffusion", "cuda", "cuda-source", "extra", "fa", "delta-databricks", "moe", "vlm", "all"]
@@ -4340,7 +4473,6 @@ dependencies = [
     { name = "nvidia-cudnn-cu12" },
     { name = "nvidia-ml-py" },
     { name = "nvidia-nvshmem-cu12", marker = "(platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (platform_machine != 'aarch64' and platform_machine != 'x86_64' and extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm') or (platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (sys_platform != 'linux' and extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
-    { name = "nvidia-resiliency-ext" },
     { name = "nvtx" },
     { name = "omegaconf" },
     { name = "pillow" },
@@ -4487,7 +4619,6 @@ requires-dist = [
     { name = "nvidia-cutlass-dsl", marker = "extra == 'vllm'", specifier = ">=4.4.0.dev1" },
     { name = "nvidia-ml-py" },
     { name = "nvidia-nvshmem-cu12", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')" },
-    { name = "nvidia-resiliency-ext" },
     { name = "nvtx" },
     { name = "omegaconf" },
     { name = "pillow", specifier = ">=11.3.0" },
@@ -4511,8 +4642,8 @@ requires-dist = [
     { name = "torchvision", marker = "sys_platform == 'darwin'", specifier = "==0.25.0", index = "https://pypi.org/simple" },
     { name = "transformer-engine", extras = ["pytorch"], marker = "extra == 'automodel'", specifier = ">=2.9.0a0,<2.12.0" },
     { name = "transformer-engine", extras = ["pytorch"], marker = "extra == 'mcore'", specifier = "==2.12.0" },
-    { name = "transformers", specifier = "==5.2.0" },
-    { name = "transformers", marker = "extra == 'automodel'", specifier = ">=5.0.0" },
+    { name = "transformers", specifier = "==5.3.0" },
+    { name = "transformers", marker = "extra == 'automodel'", specifier = ">=5.3.0" },
     { name = "triton", marker = "(platform_machine == 'aarch64' and sys_platform == 'linux') or (platform_machine == 'x86_64' and sys_platform == 'linux')", index = "https://download.pytorch.org/whl/cu129" },
     { name = "vllm", marker = "extra == 'vllm'", specifier = "==0.17.0" },
     { name = "wandb", specifier = ">=0.25.0" },
@@ -4799,14 +4930,43 @@ name = "nv-grouped-gemm"
 version = "1.1.4.post7"
 source = { git = "https://github.com/fanshiqing/grouped_gemm?tag=v1.1.4.post7#6dfaf60e6112166b8b82e9210b51c7f557956f0a" }
 dependencies = [
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "(extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra != 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra != 'extra-7-nemo-rl-fsdp' and extra != 'extra-7-nemo-rl-mcore' and extra != 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
-    { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" }, marker = "extra == 'extra-7-nemo-rl-automodel' or extra == 'extra-7-nemo-rl-mcore' or extra == 'extra-7-nemo-rl-sglang' or extra != 'extra-7-nemo-rl-vllm'" },
+    { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" } },
     { name = "setuptools" },
-    { name = "torch", version = "2.10.0", source = { registry = "https://pypi.org/simple" }, marker = "sys_platform == 'darwin' or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
-    { name = "torch", version = "2.10.0+cu129", source = { registry = "https://download.pytorch.org/whl/cu129" }, marker = "sys_platform != 'darwin' or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
+    { name = "torch", version = "2.10.0", source = { registry = "https://pypi.org/simple" }, marker = "(sys_platform == 'darwin' and extra == 'extra-7-nemo-rl-automodel') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
+    { name = "torch", version = "2.10.0+cu129", source = { registry = "https://download.pytorch.org/whl/cu129" }, marker = "(sys_platform != 'darwin' and extra == 'extra-7-nemo-rl-automodel') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
     { name = "wheel" },
 ]
 
+[[package]]
+name = "nv-one-logger-core"
+version = "2.3.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "overrides" },
+    { name = "pydantic" },
+    { name = "strenum" },
+    { name = "toml" },
+    { name = "typing-extensions" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/3b/37/963095797035f371e0db6ea761f5aaccb624fc786af217115b423baeb0e2/nv_one_logger_core-2.3.1.tar.gz", hash = "sha256:cbb2f87604c78b96a302f32d87199902129d76153a73a20f8455a250b3246c1d", size = 52640, upload-time = "2025-10-29T21:11:55.812Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ee/c4/ea91554c4fcbff66057f667690101d7a4b965605741350ac661b03fa6c46/nv_one_logger_core-2.3.1-py3-none-any.whl", hash = "sha256:0c8b77bcdac4daa1ea913bf8d4afd2a057bd5526e3654ac39f67caba157341a6", size = 63066, upload-time = "2025-10-29T21:11:52.753Z" },
+]
+
+[[package]]
+name = "nv-one-logger-training-telemetry"
+version = "2.3.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "nv-one-logger-core" },
+    { name = "strenum" },
+    { name = "typing-extensions" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/c5/21/016fa067967734d52f1ccf5a2a37a1a65216f2d7053bc2b85872cce956ca/nv_one_logger_training_telemetry-2.3.1.tar.gz", hash = "sha256:8c67940ea71799afaf1f46df3ba2f52f93aea26321c6f1c1d54aae02efc2a4af", size = 44435, upload-time = "2025-10-29T21:21:42.035Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/e5/15/97e6e4ddfe5fc35bcee74a45b7c33fb73abb83713c7dfa26420b971a86c3/nv_one_logger_training_telemetry-2.3.1-py3-none-any.whl", hash = "sha256:5319443829b59378a498c3c62ac98973e14f31be675c229ff2b14e2fe109aa0b", size = 44140, upload-time = "2025-10-29T21:21:40.72Z" },
+]
+
 [[package]]
 name = "nvdlfw-inspect"
 version = "0.2.2"
@@ -5061,22 +5221,19 @@ wheels = [
 
 [[package]]
 name = "nvidia-resiliency-ext"
-version = "0.4.1"
-source = { registry = "https://pypi.org/simple" }
+version = "0.5.0"
+source = { git = "https://github.com/NVIDIA/nvidia-resiliency-ext.git?rev=v0.5.0#5eb5f7ec84e9aa1bf45c403b06d6ef766ea6784a" }
 dependencies = [
     { name = "defusedxml" },
+    { name = "nv-one-logger-core" },
+    { name = "nv-one-logger-training-telemetry" },
     { name = "nvidia-ml-py" },
     { name = "packaging" },
     { name = "psutil" },
-    { name = "pynvml" },
     { name = "pyyaml" },
     { name = "torch", version = "2.10.0", source = { registry = "https://pypi.org/simple" }, marker = "sys_platform == 'darwin' or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
     { name = "torch", version = "2.10.0+cu129", source = { registry = "https://download.pytorch.org/whl/cu129" }, marker = "sys_platform != 'darwin' or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
 ]
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/70/05/38d491962273c7905708762279f440520eb79f3c00b67a023497215ad023/nvidia_resiliency_ext-0.4.1-cp312-cp312-manylinux_2_31_aarch64.whl", hash = "sha256:b3bd5f01535574b16d0f38bca6e39afe3806c4a2896eee1b321cd944e00025a7", size = 444570, upload-time = "2025-07-17T03:50:58.877Z" },
-    { url = "https://files.pythonhosted.org/packages/18/8b/4cb8aa2bbdf3705d3034c3f3dacdadb03b3b7dd3dc7f5200e64663fb477f/nvidia_resiliency_ext-0.4.1-cp312-cp312-manylinux_2_31_x86_64.whl", hash = "sha256:ca9f8de465af345952bedbea53c90c0e2323d88cfd830ded0e806fad91845c0e", size = 450280, upload-time = "2025-07-17T03:49:55.327Z" },
-]
 
 [[package]]
 name = "nvidia-sphinx-theme"
@@ -5568,6 +5725,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/a5/39/4c07f1d1f8e6ed85db9fe73a021113795a05aae8a84f36f0bdebb08bfde8/outlines_core-0.2.11-cp313-cp313-win_amd64.whl", hash = "sha256:ad46698564c9b13cbfbc744067de12be73bd740d7b2de20ec6b979ad7511f7c9", size = 2060567, upload-time = "2025-05-19T10:12:39.228Z" },
 ]
 
+[[package]]
+name = "overrides"
+version = "7.7.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/36/86/b585f53236dec60aba864e050778b25045f857e17f6e5ea0ae95fe80edd2/overrides-7.7.0.tar.gz", hash = "sha256:55158fa3d93b98cc75299b1e67078ad9003ca27945c76162c1c0766d6f91820a", size = 22812, upload-time = "2024-01-27T21:01:33.423Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/2c/ab/fc8290c6a4c722e5514d80f62b2dc4c4df1a68a41d1364e625c35990fcf3/overrides-7.7.0-py3-none-any.whl", hash = "sha256:c7ed9d062f78b8e4c1a7b70bd8796b35ead4d9f510227ef9c5dc7626c60d7e49", size = 17832, upload-time = "2024-01-27T21:01:31.393Z" },
+]
+
 [[package]]
 name = "packaging"
 version = "25.0"
@@ -5639,6 +5805,28 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/cc/20/ff623b09d963f88bfde16306a54e12ee5ea43e9b597108672ff3a408aad6/pathspec-0.12.1-py3-none-any.whl", hash = "sha256:a0d503e138a4c123b27490a4f7beda6a01c6f288df0e4a8b79c7eb0dc7b4cc08", size = 31191, upload-time = "2023-12-10T22:30:43.14Z" },
 ]
 
+[[package]]
+name = "peft"
+version = "0.18.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "accelerate" },
+    { name = "huggingface-hub" },
+    { name = "numpy", version = "2.4.2", source = { registry = "https://pypi.org/simple" } },
+    { name = "packaging" },
+    { name = "psutil" },
+    { name = "pyyaml" },
+    { name = "safetensors" },
+    { name = "torch", version = "2.10.0", source = { registry = "https://pypi.org/simple" }, marker = "(sys_platform == 'darwin' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
+    { name = "torch", version = "2.10.0+cu129", source = { registry = "https://download.pytorch.org/whl/cu129" }, marker = "(sys_platform != 'darwin' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-fsdp') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-mcore') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-automodel' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-fsdp' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-sglang') or (extra == 'extra-7-nemo-rl-mcore' and extra == 'extra-7-nemo-rl-vllm') or (extra == 'extra-7-nemo-rl-sglang' and extra == 'extra-7-nemo-rl-vllm')" },
+    { name = "tqdm" },
+    { name = "transformers" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/d8/48/147b3ea999560b40a34fd78724c7777aa9d18409c2250bdcaf9c4f2db7fc/peft-0.18.1.tar.gz", hash = "sha256:2dd0d6bfce936d1850e48aaddbd250941c5c02fc8ef3237cd8fd5aac35e0bae2", size = 635030, upload-time = "2026-01-09T13:08:01.136Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b3/14/b4e3f574acf349ae6f61f9c000a77f97a3b315b4bb6ad03791e79ae4a568/peft-0.18.1-py3-none-any.whl", hash = "sha256:0bf06847a3551e3019fc58c440cffc9a6b73e6e2962c95b52e224f77bbdb50f1", size = 556960, upload-time = "2026-01-09T13:07:55.865Z" },
+]
+
 [[package]]
 name = "pexpect"
 version = "4.9.0"
@@ -6509,6 +6697,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/ca/31/d4e37e9e550c2b92a9cbc2e4d0b7420a27224968580b5a447f420847c975/pytest_xdist-3.8.0-py3-none-any.whl", hash = "sha256:202ca578cfeb7370784a8c33d6d05bc6e13b4f25b5053c30a152269fd10f0b88", size = 46396, upload-time = "2025-07-01T13:30:56.632Z" },
 ]
 
+[[package]]
+name = "python-box"
+version = "6.1.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/9a/85/b02b80d74bdb95bfe491d49ad1627e9833c73d331edbe6eed0bdfe170361/python-box-6.1.0.tar.gz", hash = "sha256:6e7c243b356cb36e2c0f0e5ed7850969fede6aa812a7f501de7768996c7744d7", size = 41443, upload-time = "2022-10-29T22:30:45.515Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/88/c6/6d1e368710cb6c458ed692d179d7e101ebce80a3e640b2e74cc7ae886d6f/python_box-6.1.0-py3-none-any.whl", hash = "sha256:bdec0a5f5a17b01fc538d292602a077aa8c641fb121e1900dff0591791af80e8", size = 27277, upload-time = "2022-10-29T22:30:43.645Z" },
+]
+
 [[package]]
 name = "python-dateutil"
 version = "2.9.0.post0"
@@ -6863,6 +7060,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/97/ec/889fbc557727da0c34a33850950310240f2040f3b1955175fdb2b36a8910/requests_mock-1.12.1-py2.py3-none-any.whl", hash = "sha256:b1e37054004cdd5e56c84454cc7df12b25f90f382159087f4b6915aaeef39563", size = 27695, upload-time = "2024-03-29T03:54:27.64Z" },
 ]
 
+[[package]]
+name = "requests-toolbelt"
+version = "1.0.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "requests" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/f3/61/d7545dafb7ac2230c70d38d31cbfe4cc64f7144dc41f6e4e4b78ecd9f5bb/requests-toolbelt-1.0.0.tar.gz", hash = "sha256:7681a0a3d047012b5bdc0ee37d7f8f07ebe76ab08caeccfc3921ce23c88d5bc6", size = 206888, upload-time = "2023-05-01T04:11:33.229Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/3f/51/d4db610ef29373b879047326cbf6fa98b6c1969d6f6dc423279de2b1be2c/requests_toolbelt-1.0.0-py2.py3-none-any.whl", hash = "sha256:cccfdd665f0a24fcf4726e690f65639d272bb0637b9b92dfd91a5568ccf6bd06", size = 54481, upload-time = "2023-05-01T04:11:28.427Z" },
+]
+
 [[package]]
 name = "rich"
 version = "13.9.4"
@@ -7844,6 +8053,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/d9/52/1064f510b141bd54025f9b55105e26d1fa970b9be67ad766380a3c9b74b0/starlette-0.50.0-py3-none-any.whl", hash = "sha256:9e5391843ec9b6e472eed1365a78c8098cfceb7a74bfd4d6b1c0c0095efb3bca", size = 74033, upload-time = "2025-11-01T15:25:25.461Z" },
 ]
 
+[[package]]
+name = "strenum"
+version = "0.4.15"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/85/ad/430fb60d90e1d112a62ff57bdd1f286ec73a2a0331272febfddd21f330e1/StrEnum-0.4.15.tar.gz", hash = "sha256:878fb5ab705442070e4dd1929bb5e2249511c0bcf2b0eeacf3bcd80875c82eff", size = 23384, upload-time = "2023-06-29T22:02:58.399Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/81/69/297302c5f5f59c862faa31e6cb9a4cd74721cd1e052b38e464c5b402df8b/StrEnum-0.4.15-py3-none-any.whl", hash = "sha256:a30cda4af7cc6b5bf52c8055bc4bf4b2b6b14a93b574626da33df53cf7740659", size = 8851, upload-time = "2023-06-29T22:02:56.947Z" },
+]
+
 [[package]]
 name = "supervisor"
 version = "4.3.0"
@@ -8083,6 +8301,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/b3/46/e33a8c93907b631a99377ef4c5f817ab453d0b34f93529421f42ff559671/tokenizers-0.22.1-cp39-abi3-win_amd64.whl", hash = "sha256:65fd6e3fb11ca1e78a6a93602490f134d1fdeb13bcef99389d5102ea318ed138", size = 2674684, upload-time = "2025-09-19T09:49:24.953Z" },
 ]
 
+[[package]]
+name = "toml"
+version = "0.10.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/be/ba/1f744cdc819428fc6b5084ec34d9b30660f6f9daaf70eead706e3203ec3c/toml-0.10.2.tar.gz", hash = "sha256:b3bda1d108d5dd99f4a20d24d9c348e91c4db7ab1b749200bded2f839ccbe68f", size = 22253, upload-time = "2020-11-01T01:40:22.204Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/44/6f/7120676b6d73228c96e17f1f794d8ab046fc910d781c8d151120c3f1569e/toml-0.10.2-py2.py3-none-any.whl", hash = "sha256:806143ae5bfb6a3c6e736a764057db0e6a0e05e338b5630894a5f779cabb4f9b", size = 16588, upload-time = "2020-11-01T01:40:20.672Z" },
+]
+
 [[package]]
 name = "tomlkit"
 version = "0.13.3"
@@ -8534,7 +8761,7 @@ sdist = { url = "https://files.pythonhosted.org/packages/b0/aa/5872f0944e88b9a7a
 
 [[package]]
 name = "transformers"
-version = "5.2.0"
+version = "5.3.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "huggingface-hub" },
@@ -8546,11 +8773,11 @@ dependencies = [
     { name = "safetensors" },
     { name = "tokenizers" },
     { name = "tqdm" },
-    { name = "typer-slim" },
+    { name = "typer" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/bd/7e/8a0c57d562015e5b16c97c1f0b8e0e92ead2c7c20513225dc12c2043ba9f/transformers-5.2.0.tar.gz", hash = "sha256:0088b8b46ccc9eff1a1dca72b5d618a5ee3b1befc3e418c9512b35dea9f9a650", size = 8618176, upload-time = "2026-02-16T18:54:02.867Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/fc/1a/70e830d53ecc96ce69cfa8de38f163712d2b43ac52fbd743f39f56025c31/transformers-5.3.0.tar.gz", hash = "sha256:009555b364029da9e2946d41f1c5de9f15e6b1df46b189b7293f33a161b9c557", size = 8830831, upload-time = "2026-03-04T17:41:46.119Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/4e/93/79754b0ca486e556c2b95d4f5afc66aaf4b260694f3d6e1b51da2d036691/transformers-5.2.0-py3-none-any.whl", hash = "sha256:9ecaf243dc45bee11a7d93f8caf03746accc0cb069181bbf4ad8566c53e854b4", size = 10403304, upload-time = "2026-02-16T18:53:59.699Z" },
+    { url = "https://files.pythonhosted.org/packages/b8/88/ae8320064e32679a5429a2c9ebbc05c2bf32cefb6e076f9b07f6d685a9b4/transformers-5.3.0-py3-none-any.whl", hash = "sha256:50ac8c89c3c7033444fb3f9f53138096b997ebb70d4b5e50a2e810bf12d3d29a", size = 10661827, upload-time = "2026-03-04T17:41:42.722Z" },
 ]
 
 [[package]]
@@ -8558,16 +8785,16 @@ name = "triton"
 version = "3.6.0"
 source = { registry = "https://download.pytorch.org/whl/cu129" }
 wheels = [
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6e42d084864d12b8784736fe51a0cd05f6cc56775b25c8102c9d80c421f4e298" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6f5928e6d44c34a97bbe164cceddc0ef2007121c89ebcfba5415cf452de7ee9f" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:58d57d6796b0004076315433526fe9d4af42044d430afdee1e6cd42a76bd6d09" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0075039ff27765480083b1a109999bf27110f3542f1f9fad95f0f9065a36da79" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:29ff9ec4c96f405bf2735c531af27f8cc2215927440e52fb1fbc8014961d5d12" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a619c81f8e77116b1d10aec34d62db72daa5c0fb70a6478c9afab25edb729c52" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:20ed2f770f7217b8a994cba99e7cac37e7be735a3991344990b80ab4fea964c4" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b6d800bda666ebbe7ec2146e3f5eb271ee87f8fac724011dc7f25dee9bfb5f7f" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:89b87d2837849a9bcd3d960ad2e3dac0345c6e1359b9859e29f68083cbda9b1d" },
-    { url = "https://download.pytorch.org/whl/triton-3.6.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:621c0e5973d834507ec69dcf909f35f0b02c5df490fde100433cef83359b7eeb" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6e42d084864d12b8784736fe51a0cd05f6cc56775b25c8102c9d80c421f4e298" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6f5928e6d44c34a97bbe164cceddc0ef2007121c89ebcfba5415cf452de7ee9f" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:58d57d6796b0004076315433526fe9d4af42044d430afdee1e6cd42a76bd6d09" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0075039ff27765480083b1a109999bf27110f3542f1f9fad95f0f9065a36da79" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:29ff9ec4c96f405bf2735c531af27f8cc2215927440e52fb1fbc8014961d5d12" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a619c81f8e77116b1d10aec34d62db72daa5c0fb70a6478c9afab25edb729c52" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:20ed2f770f7217b8a994cba99e7cac37e7be735a3991344990b80ab4fea964c4" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b6d800bda666ebbe7ec2146e3f5eb271ee87f8fac724011dc7f25dee9bfb5f7f" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:89b87d2837849a9bcd3d960ad2e3dac0345c6e1359b9859e29f68083cbda9b1d" },
+    { url = "https://download-r2.pytorch.org/whl/triton-3.6.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:621c0e5973d834507ec69dcf909f35f0b02c5df490fde100433cef83359b7eeb" },
 ]
 
 [[package]]
@@ -9086,6 +9313,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/a4/f5/10b68b7b1544245097b2a1b8238f66f2fc6dcaeb24ba5d917f52bd2eed4f/wsproto-1.3.2-py3-none-any.whl", hash = "sha256:61eea322cdf56e8cc904bd3ad7573359a242ba65688716b0710a5eb12beab584", size = 24405, upload-time = "2025-11-20T18:18:00.454Z" },
 ]
 
+[[package]]
+name = "wurlitzer"
+version = "3.1.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/33/90/623f99c55c7d0727a58eb2b7dfb65cb406c561a5c2e9a95b0d6a450c473d/wurlitzer-3.1.1.tar.gz", hash = "sha256:bfb9144ab9f02487d802b9ff89dbd3fa382d08f73e12db8adc4c2fb00cd39bd9", size = 11867, upload-time = "2024-06-12T10:27:30.089Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/9a/24/93ce54550a9dd3fd996ed477f00221f215bf6da3580397fbc138d6036e2e/wurlitzer-3.1.1-py3-none-any.whl", hash = "sha256:0b2749c2cde3ef640bf314a9f94b24d929fe1ca476974719a6909dfc568c3aac", size = 8590, upload-time = "2024-06-12T10:27:28.787Z" },
+]
+
 [[package]]
 name = "xattr"
 version = "1.3.0"