Keep the table outside model mappings and Metal residency. Use bounded explicit reads, reject failed lookups before GPU consumption, and remove the runtime sidecar option and paging controls. Preserve Qwen hash and speculative state handling; cover malformed files and exact recovery after prefill, decode, and MTP read failures.
Use the structural reasoning delimiter on the server and wait for the parameter delimiter while streaming in ds4-agent. Add server and bytewise agent regressions; both fail before the fixes and pass afterward on M5 Max.
Implement qwen4exp loading, tokenizer and chat formatting, PLE sidecars,
Metal text and vision graphs, built-in MTP, and recurrent KV persistence.
Integrate the CLI, agent, server, download targets, and conversion tools.
Preserve the native Q4_1 encoder, Q4_K accumulation guard, and shared
quantizer build target. Keep exact-MTP rewind and vision-cache row-width
fixes. Persist valid logits and recurrent state at completed prefill chunks,
and reuse Qwen text-answer history when tools clients omit reasoning.
Retain exact reasoning replay and test continued checkpoints, cancellation,
cache eviction, and restart reuse with and without reasoning_content. Include
reproducible charts and measurements comparing throughput before and after
the checkpoint fixes on M3 Ultra.
Add Qwen trunk steering and prompt activation capture, including GDN layers.
Keep the MTP predictor outside the 48-row bank and preserve prompt dumps
through one-token prefill tails and multi-row speculative verification.
Recover Chrome target creation and retain its last page target. Preserve
full agent transcripts in stripped-session format when KV no longer fits,
while retaining the newer automatic compaction and continuation logic.
Add steering/MTP/capture, Chrome CDP recovery, and full-session save/reload
regressions. Adapt the intended changes from fork PRs #13 and #7.
Reconcile with antirez main bd66c40, preserving DeepSeek v4.1 model IDs,
tool syntax, session state, and Metal kernel dispatch. Exclude superseded
engine code and prototypes.
Co-authored-by: Andreas Parusel <aparuael@7170c824-6152-4cf1-985e-9503832ce673.fritz.box>
Co-authored-by: david <contact@davidberard.fr>
Co-authored-by: Michael Kuebbeler <cropduster@users.noreply.github.com>
Co-authored-by: Ogilthorp3 <bertrand@nepveu.name>
Co-authored-by: anemll <realanemll@gmail.com>
Render the existing hint marker with a white-on-teal badge and a slim left rule. Keep body text, redirected output and the model transcript unchanged.
Test every stream split, unfinished markers, quoted continuations, literal code and both tool syntaxes. Verify terminal colors and UTF-8 at multiple widths, including redraws between fragments.
Reserve summary space, recover already-full conversations by summarizing a bounded prefix, and retain every unsummarized token. Reject oversized input without corrupting the current conversation.
Resume unfinished responses in the same assistant message while preserving thinking state and the output budget. Discard incomplete tool calls before compacting and retry without executing partial arguments.
Cover real-model compaction, generated-code correctness, budget accounting, and recovery in a PTY regression; add deterministic boundary tests.
Adapt teaching-mode PR #391 to /hints on|off. Append the full
instruction once per context and short state changes at safe
assistant boundaries. New and resumed sessions start off;
compaction reinstates the current policy.
Use ordinary Markdown asides and the existing renderer, without
a learner profile, new tool, or persistent preference.
Tested on M5 Max IT with DeepSeek Flash and GLM Flash coding sessions,
prefix reuse, queued changes, compaction, save/switch/new, frontend
and PTY tests, and agent ASAN/UBSAN checks.
Closes#391
Co-authored-by: Salvatore Sanfilippo <antirez@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Replace files atomically with metadata and concurrent-change checks. Stream bounded reads with exact continuation, report incomplete searches, preserve literal [upto] edits, and avoid silently truncating user text.
Drain shell output and enforce deadlines independently of model progress. Synchronize observations and cleanup, bound log tails, honor status refresh requests, and filter terminal commands from displayed shell output. Align tool prompts and schemas with actual behavior and vision availability.
Add file, job, schema and PTY regressions. Combined fixes passed frontend tests, ASan/UBSan, ThreadSanitizer and DeepSeek/GLM editing smoke tests on M5 Max IT. Linux-specific metadata preservation still needs native QA.
Buffer complete terminal updates, repaint status without erasing live input, and retain pending redraws across throttling. Preserve fragmented UTF-8, navigation and paste input, and bound tab completion safely.
Use shared Unicode width handling for input, output and queued prompts. Preserve unmatched Markdown delimiters and publish code recoloring as one update. Add focused terminal regression tests.
Distinguish image/rendering failures from results that exceed the context budget. Preserve the transcript and return the original error. Add an invalid-image regression and run the agent suite under ASan and UBSan.
Do not HTML-unescape argument bodies. Share a narrow, reversible escape for each body closing delimiter, including its literal escaped spelling. Keep streaming fragments intact across those escapes. Preserve protocol-looking text inside GLM agent arguments until the actual argument terminator. Add round-trip, streaming and agent regressions.
Restore GLM recurrent and pooled-indexer state after rejected speculation. Invalidate unsafe DeepSeek rewinds and rebuild retained agent prefixes with their images before decoding again.
Rebuild saved TP token prefixes on both ranks. Give GLM speculative cycles an explicit command and shared token limit, with protocol version 10 requiring peers to update together.
Account for live and planned Metal GLM sessions when reserving graph and streaming memory. Add model-free state and protocol tests plus a small GPU rollback regression.
Implement GLM 5.3 Q2 and Q4 inference across Metal and CUDA, including tensor parallelism, RDMA, batching, MTP, snapshots, SSD streaming, and GLM 5.2 shared-path improvements.
Add quantization and validation tools, official quality fixtures, model downloads, regression coverage, release QA, and user documentation.
The new DS4F checkpoint struggles with anchored [upto] edits. Keep exact old/new replacement as the default and make [upto] prompting and automatic insertion an explicit --edit-upto opt-in.
Add a small ds4-agent unit-test harness for anchored [upto] edits and include it in make test so regressions in old-text matching are caught by default.
Three fixes for regressions introduced by the raw-mode restoration
commit (ce5b70b):
1. agent_prompt_yes_no_ex: handle non-blocking stdin
editor_start sets O_NONBLOCK on stdin; fgets() returned NULL
immediately (EAGAIN) causing the save prompt to exit without
waiting for user input. Temporarily switch to blocking mode
around the fgets call.
2. /quit handler: stop editor before prompting
editor_restore_terminal_layout resets the scroll region but does
not disable raw mode or restore stdin flags. Call editor_stop
first so that raw mode is off and stdin is blocking when we
prompt the user. Handle AGENT_EXIT_CANCEL by restarting the
editor instead of leaving it half-stopped.
3. linenoiseRestoreRawMode: do not corrupt orig_termios
The function unconditionally overwrote orig_termios with the
current terminal attributes via tcgetattr. While the editor is
active those are the raw-mode settings (OPOST cleared). Later,
disableRawMode restores this corrupted orig_termios, leaving the
terminal in raw mode after editor_stop. puts output then lacks
carriage returns, causing garbled display (e.g. /help).
Only update orig_termios when rawmode == 0 (child process
changed terminal to cooked).
Add shared help text across the CLI, server, agent, bench, and eval tools. Expand distributed-mode guidance, clean up endpoint naming, and use a TTY-only 256-color layout with clearer section titles, option arguments, separators, examples, and explanatory text.
antirez
Luigi Colluto
fry69
Kamran Pakseresht
Salvatore Sanfilippo
user