hmarchant/speaker-id-joint-int4

Joint SpeakerID INT4 (weight-only)

4-bit group-wise weight-only quantisation of the Joint Speaker Identifier from adobe-research/speaker-identification (Interspeech 2024).

This checkpoint is a quantised child of Adobe Research’s original Joint Speaker Identifier (FP32). The architecture and trained weights are Adobe’s; only the Linear weights were packed to INT4 (group size 64). No additional training.

Metric	Value
Precision	83.33
Δ vs FP32 parent	+4.46
F1	60.24
Accuracy	63.87
Throughput	12.09 examples/s (Apple M3 Pro, MPS, batch 2)
In-memory size	294 MB (FP32 parent: 1633 MB)

Smallest Joint checkpoint (~5.6× vs the FP32 parent). Precision rises because the model becomes more conservative; recall, F1, and accuracy drop versus the FP32/INT8 parent. Unpack-on-MPS is slower than FP16/INT8.

The INT8 sibling (hmarchant/speaker-id-joint-int8) matches the FP32 parent on precision/F1/accuracy.

Parent model

Parent	Joint Speaker Identifier (FP32), Adobe Research
Original weights	logs/mediasum-joint/best-model.mdl in adobe-research/speaker-identification
Paper	Identifying Speakers in Dialogue Transcripts: A Text-based Approach Using Pretrained Language Models (Interspeech 2024)
Backbone	FacebookAI/roberta-large
Relation	weight-only INT4 quantisation of the Adobe Joint checkpoint

Files

• model.pt — quantised bundle (scheme=int4_weight, group size 64)
• config.json — metrics and load metadata

Load

from huggingface_hub import hf_hub_download

path = hf_hub_download("hmarchant/speaker-id-joint-int4", "model.pt")

License

Derived from Adobe Research Speaker Identification. The Adobe Research License allows non-commercial research use only.