obscura/vllm
2
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2026-06-06 00:16:14 +00:00
Code Issues Packages Projects Releases Wiki Activity

[Model] Add Gemma4 Unified (encoder-free) support (#44429)

Browse Source 
Signed-off-by: Luciano Martins <lucianommartins@users.noreply.github.com>
Co-authored-by: Luciano Martins <lucianommartins@users.noreply.github.com>
This commit is contained in:
Luciano Martins
2026-06-03 16:01:39 -03:00
committed by GitHub
parent 271328e256
commit a248b45d05
14 changed files with 791 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/features/speculative_decoding/mtp.md
+5 -4
View File
@@ -24,10 +24,11 @@ vllm serve google/gemma-4-E2B-it \
--speculative-config '{"method":"mtp","model":"gg-hf-am/gemma-4-E2B-it-assistant","num_speculative_tokens":1}'
```
The E2B, E4B, 26B-A4B, and 31B Gemma 4 IT assistant checkpoints are supported
when their configuration uses `model_type: gemma4_assistant`. vLLM maps those
checkpoints to `Gemma4MTPModel` internally and wires the assistant layers to
share KV cache with the target model.
The E2B, E4B, 12B, 26B-A4B, and 31B Gemma 4 IT assistant checkpoints are supported.
Tower-based variants use `model_type: gemma4_assistant` and the encoder-free
Gemma 4 Unified variant (12B) uses `model_type: gemma4_unified_assistant`.
vLLM maps both to `Gemma4MTPModel` internally and wires the assistant layers
to share KV cache with the target model.
If an older vLLM release logs `SpeculativeConfig(method='draft_model', ...)`
for a Gemma 4 assistant checkpoint, that release is treating the assistant as a
docs/models/supported_models.md
+8 -1
View File
@@ -562,6 +562,7 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
| `Gemma3ForConditionalGeneration` | Gemma 3 | T + I<sup>E+</sup> | `google/gemma-3-4b-it`, `google/gemma-3-27b-it`, etc. | ✅︎ | ✅︎ |
| `Gemma3nForConditionalGeneration` | Gemma 3n | T + I + A | `google/gemma-3n-E2B-it`, `google/gemma-3n-E4B-it`, etc. | | |
| `Gemma4ForConditionalGeneration` | Gemma 4 | T + I<sup>+</sup> + V + A<sup>*</sup> | `google/gemma-4-E2B-it`, etc. | | ✅︎ |
| `Gemma4UnifiedForConditionalGeneration` | Gemma 4 Unified | T + I<sup>+</sup> + V + A | `google/gemma-4-12B-it`, etc. | | ✅︎ |
| `GLM4VForCausalLM`<sup>^</sup> | GLM-4V | T + I | `zai-org/glm-4v-9b`, `zai-org/cogagent-9b-20241220`, etc. | ✅︎ | ✅︎ |
| `Glm4vForConditionalGeneration` | GLM-4.1V-Thinking | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.1V-9B-Thinking`, etc. | ✅︎ | ✅︎ |
| `Glm4vMoeForConditionalGeneration` | GLM-4.5V | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5V`, etc. | ✅︎ | ✅︎ |
@@ -664,10 +665,16 @@ Some models are supported only via the [Transformers modeling backend](#transfor
For `Gemma4ForConditionalGeneration`:
- audio input is only supported by the `gemma-4-E2B` and `gemma-4-E4B` variants.
- The model does not ingest videos directly. However, vLLM’s Gemma 4 implementation supports video inputs by handling video processing internally. Users can send videos directly in the message structure to vLLM, where they are converted into text and image frames before being passed to the model.
- Gemma 4 assistant checkpoints for speculative decoding use vLLM's Gemma
- Gemma 4 assistant checkpoints for speculative decoding use vLLM’s Gemma
4 MTP path, not generic draft-model speculative decoding. See the
[Gemma 4 assistant model MTP example](../features/speculative_decoding/mtp.md#gemma-4-assistant-models).
!!! note
For `Gemma4UnifiedForConditionalGeneration`:
- This is the encoder-free Gemma 4 variant (e.g. `gemma-4-12B-it`). Unlike the tower-based `Gemma4ForConditionalGeneration`, it has **no SigLIP vision encoder** and **no audio encoder**. Raw pixel patches are projected directly into LM space via a Dense+LayerNorm pipeline with factorized positional embeddings, and raw audio waveform frames are projected directly through a multimodal embedder.
- All modalities (image, video, audio) are supported.
- Gemma 4 Unified assistant checkpoints (`model_type: gemma4_unified_assistant`) use the same MTP path as the tower-based variant. See the [Gemma 4 assistant model MTP example](../features/speculative_decoding/mtp.md#gemma-4-assistant-models).
!!! note
For `InternVLChatModel`, only InternVL2.5 with Qwen2.5 text backbone (`OpenGVLab/InternVL2.5-1B` etc.), InternVL3 and InternVL3.5 have video inputs support currently.