ThRetNet / Reference / v0.1

Reference releases

Runnable training and atomic-capability reference for the earlier three-bank model.

Package
0.1.0 · Previous
Runtime
Python 3.11+
Parameters
13,415,232
Files
21

Archive SHA-2566d4ba3d7b27ad4fd0e01bd41342a3e3f776629bf60db394960141b34be2f7c2c

Available releases

Choose a version

Why v1.3

The prompt memory splits in two

The defining change in v1.3 is the bifurcation of the prompt memory bank into reasoning-owned and finalization-owned banks. During prompt prefill, the same tokens are processed in two synchronized branches. Each subsequent phase therefore reads a prompt representation formed in its own class space, preventing conflicting class interpretations from competing inside a shared prompt bank.

Because the prompt is no longer treated as an independent compute phase, v1.3 eliminates the prompt-phase read heads and prompt-only class-phase parameter sets. Structured butterfly projections separately reduce the weight of the memory mixer. Together, these changes reduce the complete per-layer block from 999,120 to 715,520 parameters—about 28 percent.

That reduction made room to increase the MLP width from 464 to 512 and expand the network from 10 to 13 layers while reducing the representative model from about 13.4 million to 12.7 million parameters. The residual width remains 256.

Release comparison

What changed

v1.3 changes the scope of the reference package as well as the model. The older training harness and capability builders remain available in v0.1.

Areav1.3 · Latestv0.1 · Previous
Primary purposeArchitecture, correctness, and portability referenceRunnable training and atomic-capability reference
Prompt memoryReasoning-owned and finalization-owned prompt banks; four banks totalOne shared prompt bank; three banks total
Prompt phaseTwo synchronized prompt branches; no prompt-only read heads or class tablePrompt, reasoning, and finalization are independently parameterized phases
Geometry13 layers · residual width 256 · MLP width 51210 layers · residual width 256 · MLP width 464
Parameters12,716,064 in the README example configuration13,415,232 in the published package configuration
Included materialArchitecture specification, checkpoint and parity tests, portable model, isolated DGX Spark artifactTokenizer, UGI, CopyPrompt, AtomicOrder, training and generation tools

Selected release

v0.1: previous reference

This release evaluates the normalized RetNet state update literally, one token at a time. It includes the tokenizer, UGI implementation, CopyPrompt and AtomicOrder builders, and training and generation tools that are outside the narrower v1.3 correctness package.

v0.1 repository snapshot

Files

21 entries
  1. thretnet-reference-v0.1
  2. README.md

7.5 KB · MD

Download

Loading README.md