• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

opendatahub-io / kubeflow-sdk / 31783529735
86%

Build:
DEFAULT BRANCH: main
Ran 14 Aug 2026 08:22AM UTC
Jobs 1
Files 142
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

14 Aug 2026 08:20AM UTC coverage: 85.541% (-1.3%) from 86.877%
31783529735

push

github

web-flow
feat: add SpeculativeDecodingTrainer with Eagle3 draft model training support (#138)

* feat: Add SpeculativeDecodingTrainer with TRAIN_ONLY and DATA_ONLY mode support  (#123)

* feat: add SpeculativeDecodingTrainer for TRAIN_ONLY mode

Add SpeculativeDecodingTrainer supporting TRAIN_ONLY mode for training
speculative decoding draft models (Eagle3) using the speculators library.

- Training script injection via inspect.getsource()
- In-memory progression tracking with HTTP metrics endpoint
- PVC mount deduplication for shared storage
- Support for both pvc:// URIs and direct paths for output_dir
  and hidden_states_path
- Comprehensive validation and 45 unit tests

* fix: remove unused save_path from instrumentation functions

* feat: add DATA_ONLY mode support to SpeculativeDecodingTrainer

* remove num_nodes from SpeculativeDecodingTrainer

* fix: write termination message directly on phase completion for DATA_ONLY progression tracking

* fix: add target_layer_ids support, data_path validation, sidecar PVC check, and thread-safe termination

* feat: add AutoConfig layer detection, strict PVC validation, and HF_TOKEN forwarding to sidecar

* fix: disable trust_remote_code in AutoConfig.from_pretrained

* fix: pass HF_TOKEN to AutoConfig for gated model support

* fix: inject _REGEN_SCRIPT_B64 into rendered training script

* fix review comments: from_pretrained guard, collate dtype, text_config traversal, race condition, layer IDs format, seq_len default, double nesting

* fix: revert layer_ids_str to comma-separated to avoid double-nested JSON in sidecar

* fix: move distributed barrier outside exists check to prevent NCCL deadlock

* fix: account for completed epochs in progression percentage calculation

* fix: use per-epoch step with min cap for progression tracking

* fix: delete interrupted checkpoint when resume_from_checkpoint is enabled

* fix: only rank 0 deletes interrupted checkpoint to prevent multi-GPU race

* feat: add mandatory resour... (continued)

1288 of 1718 new or added lines in 8 files covered. (74.97%)

12927 of 15112 relevant lines covered (85.54%)

0.86 hits per line

Uncovered Changes

Lines Coverage ∆ File
313
49.84
kubeflow/trainer/rhai/speculator.py
107
39.2
kubeflow/trainer/rhai/instrumentation/speculator_progression.py
5
99.38
kubeflow/trainer/rhai/speculator_test.py
5
86.52
0.1% kubeflow/trainer/rhai/utils.py
Jobs
ID Job ID Ran Files Coverage
1 python-3.11 - 31783529735.1 14 Aug 2026 08:22AM UTC 142
85.54
GitHub Action Run
Source Files on build 31783529735
  • Tree
  • List 142
  • Changed 5
  • Source Changed 5
  • Coverage Changed 4
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #31783529735
  • 4e38b932 on github
  • Prev Build on main (#31782453607)
  • Delete
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc