Repository navigation
Issue forms: bug report, model request, feature request; no blank issues - #155
Merged
Merged
Conversation
Bug reports ask for the command, device, model, 1bit version, hardware and output. Questions go to Q&A, larger changes to an RFC and security reports to a private advisory (config.yml contact links). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
PR Reviewer Guide 🔍Here are some key observations to aid the review process:
|
bong-water-water-bong
pushed a commit
that referenced
this pull request
Oct 30, 2026
…epSeek-V4.1-Flash, GET_ROWS bound) third_party/llama.cpp: `d8d2f5b39f22` -> `5e6388a5a852` (1bit-MONSTER/llama.cpp 1bit/hrx-vulkan-patched). third_party/hrx-system stays at `82804197bcbe` (the commit config/hrx-long-prompt-validated-pin.config names). Fork commits new to the engine: - `fe04824` ggml-hrx: an env-gated host-gap profiler (engine #572) (#154) - `b8861a9` kv: verify the zero padding of the KV cache, and repair it (engine #528) (#155) - `9e8453a` model: DeepSeek-V4.1-Flash (deepseek41) port of upstream #31086 + #31300 (#156) - `f61f290` dsv41: drop the dead llm_arch_supports_mixed_batch definition (#157) - `01d7df4` dsv41: the deepseek41 and DSpark graphs never expanded their output (#158) - `5e6388a` ggml-hrx: GET_ROWS for a table over 262208 rows (engine #572) (#159) 44 files, +4,289/-209: a new model architecture (deepseek41), the KV zero-padding repair, the GET_ROWS row/byte bound with its matcher, the host-gap profiler, and their tests (test-backend-ops cases, test-kv-zero, test-llama-archs).
bong-water-water-bong
added a commit
that referenced
this pull request
Oct 31, 2026
* Bump HRX: llama.cpp 5e6388a5a (host-gap profiler, KV zero padding, DeepSeek-V4.1-Flash, GET_ROWS bound) third_party/llama.cpp: `d8d2f5b39f22` -> `5e6388a5a852` (1bit-MONSTER/llama.cpp 1bit/hrx-vulkan-patched). third_party/hrx-system stays at `82804197bcbe` (the commit config/hrx-long-prompt-validated-pin.config names). Fork commits new to the engine: - `fe04824` ggml-hrx: an env-gated host-gap profiler (engine #572) (#154) - `b8861a9` kv: verify the zero padding of the KV cache, and repair it (engine #528) (#155) - `9e8453a` model: DeepSeek-V4.1-Flash (deepseek41) port of upstream #31086 + #31300 (#156) - `f61f290` dsv41: drop the dead llm_arch_supports_mixed_batch definition (#157) - `01d7df4` dsv41: the deepseek41 and DSpark graphs never expanded their output (#158) - `5e6388a` ggml-hrx: GET_ROWS for a table over 262208 rows (engine #572) (#159) 44 files, +4,289/-209: a new model architecture (deepseek41), the KV zero-padding repair, the GET_ROWS row/byte bound with its matcher, the host-gap profiler, and their tests (test-backend-ops cases, test-kv-zero, test-llama-archs). * registry+census: derive the new pin's data (DeepSeek-V4.1-Flash now maps) Two generated/reviewed data sets derive from the pinned backends, and the pin move has to carry both or the tree's own checks go red at the new pin. Reported by CI on the bump head; both reproduced locally first. **registry (ctest registry_pins):** "llama.cpp (hrx) recorded at d8d2f5b39, pinned at 5e6388a5a" - `tools/registry_build.py` regenerated `registry/architectures.json` at the new pin: source string, plus `DeepseekV41ForCausalLM -> deepseek41, hrx`, counts hrx 293 -> 294. That map is legitimate at `5e6388a5a`: `conversion/__init__.py` maps the class, `src/llama-arch.cpp` registers `LLM_ARCH_DEEPSEEK41`, and `src/models/deepseek41.cpp` implements hparams/tensors/graph/engram (fork #156 = upstream #31086 + #31300). The class had been reviewed as *not an alias* of `deepseek4`, so `--check-gaps` correctly refused a silent map: `registry/significant.json` now carries a `mapped_ok` reason naming the commit and the port, its `family` text no longer claims the pinned path runs "neither", and `docs/arch-gaps.md` gets the dated note (the "not mapped" recheck it quotes was against `6b397ee9`, before the port). `tests/registry_gaps_test.py`'s committed-state invariant changes from "no reviewed class is a key in the registry" to "every reviewed class the registry maps carries a non-vacuous `mapped_ok`", with the CLI case driven by a class that genuinely has no reason. **census (ctest census_watch):** the watcher's test listed `DeepseekV41ForCausalLM` among the classes reviewed *report-only* (#522: "needs real support: engram + candidate blocks"), which is exactly what the new pin changes. The entry moves out of that table and becomes an assertion of the opposite: the same model is now COVERED, exit 0, with no `significant` alert and no `reviewed_report` line - alerting on a class the engine supports is the false alarm this watcher exists to avoid. `registry/census.json` itself needs no regeneration: it is the Hugging Face snapshot (`date`/`coverage`/`raw`, written by `tools/census.py`), not pin-derived. Green locally at this pin: `registry_build.py --check-pins`, `--check-gaps`, `tests/registry_gaps_test.py`, `tests/census_watch_test.py`. --------- Co-authored-by: bong-water-water-bong <bong-water-water-bong@1bit.gg>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds issue forms under
.github/ISSUE_TEMPLATE/:bug): what happened, the exact command, the device (dropdown of every--device), the model and where it came from,1bit version, hardware and OS, and the output. It also asks whether you searched existing issues.model request, a new label): Hugging Face link, architecture, the devices wanted, and what was tried. It points at the model registry (https://1bit.gg/registry.html) first.enhancement): for small changes only. Larger ones (backends, devices, dependencies, formats, APIs) go to an RFC, per CONTRIBUTING.md.Each form carries the Apache notice. The YAML parses, and
tools/copyright.py --checkpasses.🤖 Generated with Claude Code