Adding Model Support

Adding New Model Support in Megatron-Bridge

Phase 1: Discovery

Step 1 — Get the HF model link

Ask the user for the HuggingFace model link (e.g. https://huggingface.co/Qwen/Qwen3.5-VL-27B).

If the model is not public, ask the user to provide the config.json file directly.

Step 2 — Fetch and analyze config.json

Read the model's config.json from HuggingFace (or from the user-provided file). Key fields to extract:

• model_type — used for @register_bridge(model_type=...)
• architectures — the HF model class name (used for source=... in registration)
• tie_word_embeddings — critical for weight tying
• Architecture fields: num_hidden_layers, hidden_size, intermediate_size, num_attention_heads, num_key_value_heads, vocab_size, max_position_embeddings, rope_theta, etc.
• MoE fields (if present): num_local_experts, num_experts_per_tok, moe_intermediate_size
• MLA fields (if present): q_lora_rank, kv_lora_rank, qk_nope_head_dim, qk_rope_head_dim

If there are config fields you don't recognize from previously supported models (check CONFIG_MAPPING in model_bridge.py and existing bridges), this likely indicates a new architectural block (e.g., a novel attention variant, custom normalization, or a new layer type). Ask the user to provide the HuggingFace modeling_*.py implementation of that block so you can understand the computation and create the correct Megatron-side mapping or custom module.

Step 3 — Determine VLM vs LLM

VLM (Vision-Language Model) if config.json contains:

• text_config AND vision_config sub-configs
• Note: VLMs may or may not have "VL" in the name

LLM (Text-only) if:

• No text_config / vision_config
• Single flat config for the language model

This distinction affects:

• Which files to create (VLMs need a model.py combining vision + language)
• Where to read config fields from (text_config vs top-level for VLMs)
• Test patterns (VLMs need vision inputs in functional tests)

Step 4 — Check for quantized weights (FP8 / FP4)

Inspect the HF checkpoint's model.safetensors (or model.safetensors.index.json) for quantized weight dtypes such as float8_e4m3fn (FP8) or uint8/uint4 with accompanying *_scale_inv or *_scale tensors. Common signs:

• config.json mentions quantization_config or dtype fields like "torch_dtype": "float8_e4m3fn"
• Safetensors contain weight_scale_inv keys alongside the main weight keys
• The model card mentions FP8/FP4/INT4 weights

Why this matters: The bridge's import_ckpt path does not automatically dequantize — it loads raw quantized values as-is. This produces a silently broken model (random-level loss, huge grad norms) instead of raising an error.

Fix: Dequantize before or during conversion. The current in-repo pattern is to use a bridge hook plus the shared helpers in src/megatron/bridge/models/conversion/quantization_utils.py. Existing examples include src/megatron/bridge/models/ministral3/ministral3_bridge.py, src/megatron/bridge/models/deepseek/deepseek_v3_bridge.py, and src/megatron/bridge/models/minimax_m2/minimax_m2_bridge.py.

Override maybe_modify_loaded_hf_weight() in the bridge class to dequantize on the fly during import:

def maybe_modify_loaded_hf_weight(self, hf_param, hf_state_dict):
    weight = hf_state_dict[hf_param]
    scale_key = hf_param + "_scale_inv"
    if weight.dtype == torch.float8_e4m3fn and scale_key in hf_state_dict:
        return weight.to(torch.bfloat16) * hf_state_dict[scale_key].to(torch.bfloat16)
    return weight

Always add a sanity check in the verification workflow (e.g., print std of a weight tensor — quantized models typically have std ≈ 13 before dequantization vs std ≈ 0.006 after). Also add or update focused tests when touching export/import quantization paths; see tests/unit_tests/models/test_fp8_param_export.py for current FP8 export coverage.

Phase 2: Bridge Support

File structure

LLM — Reference: Qwen2 (src/megatron/bridge/models/qwen/qwen2_bridge.py)

src/megatron/bridge/models/<model>/
├── __init__.py
├── <model>_bridge.py      # Config + weight mappings (no provider file needed)
└── modeling_<model>/      # (optional) Custom nn.Module implementations if needed
    └── ...

VLM — Reference: Qwen3.5-VL (src/megatron/bridge/models/qwen_vl/)

src/megatron/bridge/models/<model>/
├── __init__.py
├── <model>_bridge.py         # Config + weight mappings
├── <model>_provider.py       # Only for VLMs that need custom provide()
└── modeling_<model>/         # If using Megatron vision encoder
    ├── __init__.py
    └── model.py              # Combines vision + language

OR with HF vision encoder (Reference: Gemma3-VL):

src/megatron/bridge/models/<model>/
├── __init__.py
├── <model>_bridge.py
├── <model>_provider.py       # Only for VLMs that need custom provide()
└── modeling_<model>.py       # HF vision + Megatron language wrapper

Model-specific modeling code: If the model requires custom nn.Module implementations (e.g. a custom RoPE variant, non-standard transformer config, custom thinker/talker architecture), place them in a modeling_<model>/ directory or a single modeling_<model>.py file inside the model family folder. Use a directory when there are multiple files (model, transformer config, custom ops); use a single file when one module suffices. Never put model-specific modeling code in shared directories or as loose files in the bridge family directory — keep them namespaced under the modeling_<model> prefix.

Implementation order

LLM:

1. Bridge only — Register bridge, implement provider_bridge() and mapping_registry(). The bridge calls super().provider_bridge() to get a GPTModelProvider from CONFIG_MAPPING, then sets model-specific attributes on it. Do not create a provider file — the stock provider returned by super().provider_bridge() is usually sufficient for LLMs (e.g., GPTModelProvider, or another base provider selected via PROVIDER_CLASS). Do not add size-specific provider classes whose names combine ModelProvider with a model-size suffix. Examples of forbidden suffixes include 7B, 200M, and A3B. Model size and architecture fields should come from the Hugging Face config through AutoBridge / MegatronModelBridge config mapping. If a recipe needs a fixed architecture, configure the base provider inside the recipe function instead of exporting a provider subclass.

VLM:

1. Bridge — Register bridge, implement config and weight mappings.
2. Provider (when needed) — Only VLMs that require a custom provide() to instantiate a combined vision+language model need a provider subclass. The bridge manually calls hf_config_to_provider_kwargs(text_config) and instantiates the custom provider.
3. Model class — Combine vision encoder + language decoder.

For detailed patterns, see:

• VLM: @skills/adding-model-support/vlm-patterns.md
• LLM: @skills/adding-model-support/llm-patterns.md

Critical: tie_word_embeddings for VLMs

For VLMs, tie_word_embeddings lives on the top-level HF config, NOT on text_config. Always read from the parent config:

provider.share_embeddings_and_output_weights = getattr(hf_config, "tie_word_embeddings", False)

Critical: Config field location for VLMs

When reading HF config for VLMs, check whether each field is in:

• hf_config (top-level) — e.g. tie_word_embeddings, image_token_id, video_token_id
• hf_config.text_config — e.g. num_hidden_layers, hidden_size, etc.
• hf_config.vision_config — e.g. vision encoder dimensions

Encapsulating model-specific layers

When a new model introduces custom or non-standard layers (novel attention variants, custom normalization, fused expert layouts, MTP heads, etc.), keep all model-specific logic inside the model family directory. Do not modify shared files in src/megatron/bridge/models/conversion/ (e.g. param_mapping.py, model_bridge.py, quant_mapping.py) unless the change is genuinely reusable across multiple model families.

Principle: The bridge and provider files for a model family are your primary extension surface. Shared conversion infrastructure provides hooks and base classes — subclass them locally rather than adding conditionals to shared code.

Prefer Transformer Engine normalization

Before implementing a model-specific normalization with raw PyTorch operations, inspect the exact Transformer Engine revision in pyproject.toml/uv.lock, its upstream module API, and Megatron-Core's TENorm adapter. Production Bridge model code should assume Transformer Engine is installed. Use TENorm or the corresponding TE module directly for LayerNorm and RMSNorm; do not add model-local norm wrappers, copy an existing norm implementation, gate the norm on HAVE_TE, or provide a torch.nn fallback. Preserve eps, parameter dtype/device initialization, sequence-parallel metadata, and zero_centered_gamma; standard-gamma and zero-centered-gamma norms are not interchangeable. The attention/linear backend and normalization choice are independent, so a local attention spec should still use TE normalization.

Treat a custom norm implementation as an architectural exception, not a compatibility pattern. It is allowed only when the pinned TE API cannot represent the model's exact mathematical and parameter contract, such as a genuinely parameter-free RMSNorm while TE requires an affine weight. Prefer a framework-native module such as torch.nn.RMSNorm(elementwise_affine=False) or a direct functional call over a model-local wrapper or copied formula. Document the missing TE capability and test that the exception introduces no checkpoint parameter. Do not create a wrapper or fallback for ordinary affine normalization.

Do not introduce a dummy affine parameter merely to reach a fused kernel. For example, current TE RMSNorm always owns a learnable weight, so a genuinely weightless/scaleless RMSNorm must retain a parameter-free implementation until TE exposes matching semantics. Adding an all-ones frozen weight would change state-dict keys, optimizer state, distributed-checkpoint schema, and conversion coverage.

TE and reference framework kernels can differ numerically even when their architecture and weights match. Keep exact HF↔Megatron weight round-trip as the conversion gate, then assess forward behavior with the correlation criteria in the parity-testing skill rather than reverting to a slower norm only to reproduce one framework's operation ordering.

Strategy 1: Create a local mapping subclass

If the model has a layer whose weight layout doesn't match any existing mapping class, create a private mapping class in the bridge file or a <model>_mappings.py file in the family directory.

Example — GLM's fused expert down-projection disables grouped-export transpose:

# src/megatron/bridge/models/glm/glm_moe_mappings.py
class GLMExpertDownProjMapping(FusedExpertMapping):
    def __init__(self, megatron_param, hf_param, permute_dims=None):
        super().__init__(megatron_param, hf_param, permute_dims, transpose_on_export=False)

Example — Nemotron-H's MTP layers flatten indices during resolve:

# Inside nemotron_h_bridge.py (private to the module)
class _MTPFlatteningMapping(MegatronParamMapping):
    def resolve(self, captures):
        return AutoMapping(self._flatten(captures), ...)

Example — MiniMax-M2's non-standard QK norm layout:

# Inside minimax_m2_bridge.py (private to the module)
class _FullDimQKNormMapping(MegatronParamMapping):
    def hf_to_megatron(self, hf_weights):
        # Custom scatter logic for full-dim QK norm
        ...
    def megatron_to_hf(self, megatron_weights):
        # Custom gather logic
        ...

Strategy 2: Override bridge hooks

MegatronModelBridge provides several override hooks — use them instead of modifying the base class:

Accessing HF config in mapping_registry(): The bridge instance has self.hf_config available during conversion — it is set automatically by the dispatch system before mapping_registry() is called. Use it when your mapping registry needs config-dependent logic (e.g. dynamic MTP layer count, number of experts):

def mapping_registry(self) -> MegatronMappingRegistry:
    hf_config = getattr(self, "hf_config", None)
    num_mtp_layers = getattr(hf_config, "num_nextn_predict_layers", 0) if hf_config else 0
    ...

Do not override build_conversion_tasks() to stash self._hf_config — that pattern is deprecated.

Strategy 3: Custom provider subclass (VLMs only)

Most models do not need a provider file — the stock provider (e.g., GPTModelProvider, or another base selected via PROVIDER_CLASS) is usually sufficient for LLMs. Only create a provider subclass when a VLM needs custom provide() logic to instantiate a combined vision+language model:

# src/megatron/bridge/models/<model>/<model>_provider.py
class MyVLModelProvider(GPTModelProvider):
    image_token_id: int = 0

    def provide(self, ...):
        # Custom model construction combining vision encoder + language decoder
        ...

The bridge then references it via PROVIDER_CLASS = MyVLModelProvider or instantiates it directly in provider_bridge().

When shared file changes ARE justified

Modify param_mapping.py or model_bridge.py only when the pattern is reusable by 2+ model families. Examples of justified shared changes:

• FusedExpertMapping / FusedGatedExpertMapping — used by GLM, DeepSeek, OLMoE, etc.
• RMSNorm2ZeroCenteredRMSNormMapping — used by Gemma, Nemotron, etc.
• New CONFIG_MAPPING entries — when a standard HF config key maps to a standard provider attribute

If you're tempted to add a model-specific if model_type == "..." branch in shared code, or pattern-matching on specific weight names in shared conversion logic, that's a signal to use a local subclass or hook override instead.

Update FLOPs calculator for new architectural blocks

If the model introduces a new computational block that differs from standard attention or MLP (e.g., Gated DeltaNet / GDN linear attention, Multi-Token Prediction / MTP heads, Mamba SSM layers), update the FLOPs calculator in src/megatron/bridge/training/utils/flop_utils.py so that training throughput metrics (TFLOPs/GPU) are accurate.

When to update: Any time the new block has different FLOPs-per-token than standard self-attention or standard MLP. Common cases:

• Linear attention variants (GDN, RetNet, RWKV) — replace the O(s²) attention term with the block's actual operation count
• MTP / speculative decoding heads — add FLOPs for the extra projection and norm layers
• SSM layers (Mamba) — different recurrence FLOPs than attention
• Novel MoE routing — may change the effective expert count

How to update:

1. Read the existing transformer_flops() function in flop_utils.py to understand the structure.
2. Add a conditional block gated on a config attribute (e.g., experimental_attention_variant, mtp_num_layers). Follow the existing MoE pattern for config validation — raise on invalid types, assert list lengths, and use direct attribute access instead of getattr with fallback defaults so that misconfigurations fail explicitly.
3. Compute the per-layer FLOPs for the new block and blend it with the standard attention term based on the layer pattern.
4. Add unit tests in tests/unit_tests/training/utils/test_flop_utils.py that verify:
   • New-block FLOPs differ from pure-attention baseline
   • Exact formula matches hand-computed expected values
   • Varying the block ratio (e.g., linear_attention_freq) changes FLOPs

Reference PR: #2925 — GDN FLOPs calculator adds GDN support with both the calculator code and comprehensive tests.

Phase 3: Recipe Support

Recipes provide pre-configured training settings for each model size.

LLM recipes: src/megatron/bridge/recipes/<family>/<model>.py VLM recipes: src/megatron/bridge/recipes/<family>/<model>.py

Each recipe file defines functions for each model size + training mode:

• <model>_<size>_sft_config() — Full supervised fine-tuning
• <model>_<size>_peft_config() — LoRA/DoRA parameter-efficient fine-tuning
• <model>_<size>_pretrain_config() — Pretraining (LLM only, usually)

For detailed recipe patterns, see @skills/adding-model-support/recipe-patterns.md.

Recipes are the right API surface for model-size presets. Do not create or export size-specific provider subclasses for recipes; either call AutoBridge.from_hf_pretrained(...).to_megatron_provider(load_weights=False) to derive the provider from HF config, or instantiate the base provider class with explicit architecture fields inside the recipe function.

Export checklist

1. Family __init__.py — import and add to __all__
2. Top-level src/megatron/bridge/recipes/__init__.py — wildcard import
3. train_any_basic.py — add to config_map, docstring, and --model choices

Phase 4: Tests

Unit tests (no GPU)

tests/unit_tests/models/<model>/
├── __init__.py
├── test_<model>_bridge.py    # Mock HF config → verify provider mapping
└── test_<model>_provider.py  # (optional) Only if custom provider subclass exists

Functional tests (GPU)

tests/functional_tests/test_groups/models/<model>/
├── __init__.py
├── test_<model>_conversion.py  # Toy model HF↔Megatron roundtrip
└── test_<model>_provider.py    # compare_provider_configs (optional)

For detailed test patterns, see @skills/adding-model-support/tests-and-examples.md.

Phase 5: Docs and Examples

Examples

Model examples: examples/models/<family>/<model>/

examples/models/<family>/<model>/
├── README.md
├── inference.sh         # Generation commands (real model, reasonable output)
├── slurm_sft.sh         # SFT training on SLURM
└── slurm_peft.sh        # PEFT training on SLURM

Key deliverable requirement: The README must include working import and export commands for a real published model (e.g. Qwen/Qwen3-8B, not a toy) using scripts/conversion/convert.sh. Add a model-specific conversion wrapper only when the model requires preparation or verification that the shared CLI cannot express. The inference script must produce reasonable output — for LLMs a coherent text continuation, for VLMs a plausible image description. This is the acceptance bar: conversion runs cleanly and generation makes sense.

Documentation

Add a model page at docs/models/<type>/<model>.md covering:

• Supported variants and sizes
• Conversion commands
• Training examples (SFT, PEFT)
• Known limitations

Verification Workflow

After implementing bridge support, prompt the user to run these commands on the cluster:

1. Smoke test (single GPU)

uv run python -c "
from megatron.bridge import AutoBridge
bridge = AutoBridge.from_hf_pretrained('<org>/<model>')
provider = bridge.to_megatron_provider()
provider.tensor_model_parallel_size = 1
provider.pipeline_model_parallel_size = 1
provider.finalize()
model = provider.provide_distributed_model(wrap_with_ddp=False)
bridge.load_hf_weights(model)
for i, (name, tensor) in enumerate(bridge.export_hf_weights(model, cpu=True)):
    print(name, tuple(tensor.shape))
    if i > 10: break
"

2. Conversion roundtrip (multi-GPU)

./scripts/conversion/convert.sh import \
    --hf-model <org>/<model> \
    --megatron-path /workspace/<model> \
    --torch-dtype bfloat16

./scripts/conversion/convert.sh export \
    --hf-model <org>/<model> \
    --megatron-path /workspace/<model>/iter_0000000 \
    --hf-path /workspace/<model>-hf-export

3. Generation test

For LLMs:

uv run python examples/conversion/hf_to_megatron_generate_text.py \
    --hf_model_path <org>/<model> --prompt "Hello"

For VLMs:

uv run python examples/conversion/hf_to_megatron_generate_vlm.py \
    --hf_model_path <org>/<model> \
    --image_path "https://example.com/image.jpeg" \
    --prompt "Describe this image."

4. Run tests

uv run python -m pytest tests/unit_tests/models/<model>/ -v
uv run python -m pytest tests/functional_tests/test_groups/models/<model>/ -v --run-gpu

Quick Decision Tree

User wants to add a model
│
├─ Has HF link? ─── No ──→ Ask for link (or config.json if private)
│
├─ Has text_config + vision_config? ─── Yes ──→ VLM path
│   ├─ Has Megatron vision encoder? ──→ Megatron encoder (Qwen3.5 pattern)
│   └─ No Megatron encoder ──→ HF encoder (Gemma3 pattern)
│
└─ No vision config ──→ LLM path (bridge only, no provider file)
    ├─ Standard GPT-style? ──→ Bridge with stock mappings
    └─ Custom layers? ──→ Bridge + local mapping subclasses / hook overrides
        ├─ Custom weight layout? ──→ Local mapping subclass in family dir
        └─ Custom import/export? ──→ Override bridge hooks (maybe_modify_*)