Upload README.md

README.md

@@ -0,0 +1,48 @@
+---
+license: mit
+library_name: pytorch
+tags:
+  - co-speech-gesture-generation
+  - motion-generation
+  - smpl-x
+  - streamtalk
+datasets:
+  - H-Liu1997/BEAT2
+---
+
+# StreamTalk retrained CFG checkpoints
+
+These are hardware-agnostic FP32 generator state dictionaries for
+`DiffusionDITNetPartsFixedExpressions2PostNormInteraction2`. They were retrained
+on BEAT2 English with classifier-free condition dropout and selected with the
+released StreamTalk generation protocol plus the official EMAGE/PantoMatrix
+AESK FGD metric.
+
+| File | Intended role | Epoch | CFG | Speaker2 FGD | All FGD |
+| --- | --- | ---: | ---: | ---: | ---: |
+| `streamtalk_speaker2_combined_e0946_cfg3.pt` | Speaker2 and combined | 946 | 3 | 0.378879 | 0.250287 |
+| `streamtalk_speaker_all_e0940_cfg3.pt` | All speakers | 940 | 3 | 0.424477 | 0.217672 |
+
+The checkpoints contain 407 CPU FP32 tensors and 71,167,501 parameters. Both
+load strictly into the originally released StreamTalk model schema. H200,
+BF16, CUDA Graphs, and four-GPU execution were training/evaluation
+infrastructure choices and are not encoded in the checkpoint files.
+
+The reported values are accelerated EMAGE/PantoMatrix AESK FGD measurements,
+not paper-exact oracle measurements. A 1e-3 absolute FGD reporting tolerance
+was selected for engineering comparisons, but the final CFG=3 values have not
+been accepted against a direct B=1/M=1/full-window oracle. BC and DIV have not
+been re-evaluated for these retrained checkpoints.
+
+Verify downloads before loading:
+
+```bash
+python tools/verify_pretrained.py --weights-dir checkpoints/pretrained
+```
+
+PyTorch checkpoints use pickle internally. Only load files downloaded from the
+linked StreamTalk release and matching the published SHA256 values.
+
+End-to-end inference also requires WavLM Large, the bundled
+`SimpleSpeechModel`, SMPL-X neutral, and the BEAT2 retrieval database. Those
+runtime assets are independent of the generator checkpoint format.