Skip to content

Onboard google/gemma-4-31b-it text configuration - #296

Open
35Upesh wants to merge 2 commits into
torch-spyre:mainfrom
35Upesh:google/gemma-4-31b-it-text
Open

Onboard google/gemma-4-31b-it text configuration#296
35Upesh wants to merge 2 commits into
torch-spyre:mainfrom
35Upesh:google/gemma-4-31b-it-text

Conversation

@35Upesh

@35Upesh 35Upesh commented Aug 6, 2026

Copy link
Copy Markdown
Contributor

Summary

Register google/gemma-4-31b-it (Gemma 4 31B Instruct) as a verified causal LM checkpoint for Spyre.

Changes

  • Added google/gemma-4-31b-it to MODEL_PATH_TO_TORCH_DTYPE with torch.bfloat16
  • Registered gemma4_31b_it in CAUSAL_LM_MODELS
  • Updated ARCHITECTURE.md and README.md to reflect 47 verified checkpoints

Validation

  • ✅ CPU tests passed (17/18, with 1 unrelated pre-existing failure)
  • ✅ Spyre load and E2E tests passed
  • ✅ Token parity verified (5/5 top-1 match) against Hugging Face CPU inference

Signed-off-by: 35Upesh <35upesh@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant