Skip to content

Reasoning Budget #421

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
will-lms merged 4 commits into from
Jul 14, 2026
Merged

Reasoning Budget #421

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
75f06d3
Add reasoning budget message localization
will-lms Jul 14, 2026
c234141
Add reasoning budget localization
will-lms Jul 14, 2026
c14222e
Align reasoning budget help text
will-lms Jul 14, 2026
cb86dde
Remove unused reasoning budget info text
will-lms Jul 14, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions en/config.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -59,6 +59,8 @@
"llm.prediction.maxPredictedTokens/info": "Control the max length of the chatbot's response. Turn on to set a limit on the max length of a response, or turn off to let the chatbot decide when to stop.",
"llm.prediction.maxPredictedTokens/inputLabel": "Maximum response length (tokens)",
"llm.prediction.maxPredictedTokens/wordEstimate": "About {{maxWords}} words",
"llm.prediction.reasoning.budgetTokens/title": "Reasoning Budget",
"llm.prediction.reasoning.budgetTokens/subTitle": "Token budget for thinking: off for unrestricted, 0 for immediate end, N > 0 for token budget",
"llm.prediction.repeatPenalty/title": "Repeat Penalty",
"llm.prediction.repeatPenalty/subTitle": "How much to discourage repeating the same token",
"llm.prediction.repeatPenalty/info": "From llama.cpp help docs: \"Helps prevent the model from generating repetitive or monotonous text.\n\nA higher value (e.g., 1.5) will penalize repetitions more strongly, while a lower value (e.g., 0.9) will be more lenient.\" • The default value is <{{dynamicValue}}>",
Expand Down Expand Up @@ -134,6 +136,8 @@
"llm.load.useUnifiedKvCache/subTitle": "Controls whether concurrent predictions share a single KV cache, saving memory. Disabling this ensures each prediction can utilize the full context length, at the cost of using more memory",
"llm.load.llama.contextCheckpoints/title": "Context Checkpoints",
"llm.load.llama.contextCheckpoints/subTitle": "Maximum number of context checkpoints to keep per slot. Lower values reduce RAM usage for SWA/recurrent models. 0 disables checkpoints.",
"llm.load.llama.reasoningBudgetMessage/title": "Reasoning Budget Message",
"llm.load.llama.reasoningBudgetMessage/subTitle": "Message injected before the end-of-thinking tag when reasoning budget is exhausted (default: none)",
"load.gpuStrictVramCap/title": "Limit Model Offload to Dedicated GPU Memory",
"load.gpuStrictVramCap.customSubTitleOff": "OFF: Allow model weights to offload to shared memory if dedicated GPU memory is full",
"load.gpuStrictVramCap.customSubTitleOn": "ON: The system will limit offload of model weights to dedicated GPU memory and RAM only. Context may still use shared memory",
Expand Down Expand Up @@ -563,6 +567,7 @@
"llamaKvCacheQuantizationWarning": "KV Cache Quantization is an experimental feature that may cause issues with some models. Flash Attention must be enabled for V cache quantization. If you encounter problems, reset to the default \"F16\".",

"seedUncheckedHint": "Random Seed",
"reasoningBudgetUncheckedHint": "Unrestricted",
"ropeFrequencyBaseUncheckedHint": "Auto",
"ropeFrequencyScaleUncheckedHint": "Auto",

Expand Down
Loading