GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
dockhardman
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
dockhardman
Model Tree
google/gemma-4-E2B-it
Input Modalities
Output Modalities
Supported Functionality
All in one adapter, bilingual (Chinese / English).
Base Gemma is turn-based: you send a whole message, it generates a whole reply, done. There is no notion of time — nothing happens "while you are still talking."
This adapter makes the same model frame-synchronous. Time is sliced into 80ms frames, and on every single frame the model does two things at once: it reads the latest audio and decides what to emit. Each frame is 2 audio tokens + 1 text token, and that one text token is a live decision:
markdown
time → frame 1 frame 2 frame 3 … frame k frame k+1 …audio [you ...... speaking .............. pause] (silence)model <wait> <wait> <wait> … <wait> "Sure," "here's"…└─ stays quiet while you talk ─┘ └─ opens up on its own ─┘
<wait> = "keep listening." Emitting a word = "I'm speaking now."<wait> to words?<wait> within a few frames.Because every decision lives on a shared clock with the audio, these behaviors are the model's own, learned end-to-end — not bolted on by a separate turn-detector.
The LoRA sits only on the language model's attention/MLP projections; the audio encoder and embeddings stay frozen — we teach behavior, not perception.
The key finding behind this release: opening up, shutting up (on barge-in), and staying in a
multi-turn conversation are the same problem — the "act now" signal is a tiny fraction of the
supervised positions and gets drowned out by all the <wait> frames. Amplifying the training loss
on those few decisive positions cures all three at once, with plain supervised fine-tuning.
Bare model, all runtime heuristics off (no VAD, no nudge) — this is the model on its own:
| Capability | Metric | Result |
|---|---|---|
| Opening (standard) | replies after you finish | 11/12 |
| Opening (rambling questions) | " | 12/12 |
| Barge-in | chars still spoken after interrupt (median) | 7 |
| Multi-turn | 5-turn survival (turn 1 / overall) | 8/8 · 5/8 stable |
| Text channel | replies to typed input | 24/24 |
| Tool calling | correct call + spoken summary |
With a light runtime layer (energy-gated barge, silence-freeze), the live experience is smoother still — see the demo repo.
This is a PEFT LoRA adapter for google/gemma-4-E2B-it.
python
from peft import PeftModelfrom transformers import Gemma4ForConditionalGeneration, AutoProcessorbase = Gemma4ForConditionalGeneration.from_pretrained("google/gemma-4-E2B-it")model = PeftModel.from_pretrained(base, "dockhardman/gemma-4-E2B-duplex")processor = AutoProcessor.from_pretrained("google/gemma-4-E2B-it")
The adapter runs frame-synchronously (80ms frames), so it needs a streaming driver to feed audio and drain text/tool events. A companion demo repo with a ready-to-run web app (microphone UI
gemma-4-E2B-it base).google/gemma-4-E2B-itbibtex
@misc{gemma4e2bduplex2026,author = {allen2c},title = {Gemma-4-E2B-Duplex: a frame-synchronous full-duplex speech LoRA},year = {2026},howpublished = {\url{https://huggingface.co/dockhardman/gemma-4-E2B-duplex}}}
| 11/12 |