@qvac/llm-llamacpp 0.60.0 → 0.61.0

package/NOTICE

@@ -239,4 +239,5 @@ C++ Dependencies
     https://github.com/nlohmann/json
   qvac-fabric
     https://github.com/tetherto/qvac-fabric-llm.cpp
-
+  stb_image v2.30 (MIT or public domain)
+    https://github.com/nothings/stb

package/README.md

@@ -193,6 +193,7 @@ const config = {
 | frequency_penalty | float                                       | 0                            | Frequency penalty for sampling                        |
 | tools             | `"true"` or `"false"`                       | `"false"`                    | Enable tool calling with jinja templating             |
 | verbosity         | 0 – 3 (0=ERROR, 1=WARNING, 2=INFO, 3=DEBUG) | 0                            | Logging verbosity level                               |
+| image-max-megapixels | positive integer                            | 50                           | Maximum reported image pixel count in millions, checked before decode. Alias: `image_max_megapixels`. |
 | main-gpu          | integer, `"integrated"`, or `"dedicated"`   | —                            | GPU selection for multi-GPU systems                   |
 | backend           | comma-separated list of `cuda`, `vulkan`, `metal`, `opencl`, or `auto` | N/A | Overrides which GPU backend is used, in priority order (e.g. `"cuda,vulkan"`). `auto` means no preference. An unrecognised name is rejected; a recognised one with no device present is skipped. Use `device: "cpu"` to run on CPU. Cannot be combined with `split-mode` or `devices` |
 | split-mode        | `"none"`, `"layer"`, or `"tensor"` | `"none"`                     | How to split the model across GPUs. `"tensor"` is EXPERIMENTAL and desktop-only; `"row"` is rejected ([details](./docs/multi-gpu.md)) |
@@ -411,7 +412,7 @@ Cancelling a queued job resolves immediately, whether or not the slots it was wa
 ```js
 const LlmLlamacpp = require('@qvac/llm-llamacpp')
 
-const fit = LlmLlamacpp.assessFit({
+const fit = await LlmLlamacpp.assessFit({
   modelPath: '/models/model.gguf',
   config: { device: 'gpu', 'ctx-size': '4096' }
 })
@@ -431,7 +432,9 @@ Pinning `gpu-layers` fixes the placement: the fitter refuses to move layers off
 
 `minCtxSize` sets a floor the fitter may not reduce the context below, defaulting to llama's own `fit-ctx` of 4096. `marginBytes` is the memory to leave free on every device, defaulting to llama's `fit-target` of 1 GiB per device. `backendsDir` is where the dynamically-loaded ggml backends live.
 
-A model the fitter cannot read is `status: "error"`; only a broken request throws.
+`assessFit` returns a Promise. Concurrent calls run one at a time within this addon.
+
+A model the fitter cannot read resolves to `status: "error"`; only a broken request rejects.
 
 ## Fine-tuning
 

package/fit.d.ts

@@ -38,16 +38,15 @@ export interface LlamaFitResult {
  * and never weight data. The registry's weightless copy of a model answers the
  * same as the model itself, so this can run before anything is downloaded.
  *
- * A model the fitter cannot read comes back as `status: "error"`; only a broken
- * request throws.
+ * A model the fitter cannot read resolves to `status: "error"`; only a broken
+ * request rejects.
  *
- * Runs synchronously on the calling thread and builds the model twice without
- * allocating weights, so a caller that must stay responsive runs it off its
- * own loop.
+ * Builds the model twice without allocating weights. Concurrent calls run one
+ * at a time.
  *
  * The projection reads the model's vocabulary, and llama asserts on a
  * vocabulary it finds inconsistent. An assert aborts rather than throwing, so
  * a corrupt file ends the process instead of returning a status: take the
  * model from a source you trust, or call this behind a process boundary.
  */
-export declare function assessFit(request: LlamaFitRequest): LlamaFitResult;
+export declare function assessFit(request: LlamaFitRequest): Promise<LlamaFitResult>;

package/fit.js

@@ -6,20 +6,19 @@ exports.assessFit = assessFit;
  * and never weight data. The registry's weightless copy of a model answers the
  * same as the model itself, so this can run before anything is downloaded.
  *
- * A model the fitter cannot read comes back as `status: "error"`; only a broken
- * request throws.
+ * A model the fitter cannot read resolves to `status: "error"`; only a broken
+ * request rejects.
  *
- * Runs synchronously on the calling thread and builds the model twice without
- * allocating weights, so a caller that must stay responsive runs it off its
- * own loop.
+ * Builds the model twice without allocating weights. Concurrent calls run one
+ * at a time.
  *
  * The projection reads the model's vocabulary, and llama asserts on a
  * vocabulary it finds inconsistent. An assert aborts rather than throwing, so
  * a corrupt file ends the process instead of returning a status: take the
  * model from a source you trust, or call this behind a process boundary.
  */
-function assessFit(request) {
+async function assessFit(request) {
     // eslint-disable-next-line @typescript-eslint/no-require-imports -- native binding is resolved lazily from package prebuilds.
     const binding = require('./binding.js');
-    return binding.assessFit({ ...request, config: { ...request.config } });
+    return await binding.assessFit({ ...request, config: { ...request.config } });
 }

package/index.d.ts

@@ -244,6 +244,10 @@ declare namespace LlmLlamacpp {
         devices?: string;
         "cache-type-k"?: string;
         "cache-type-v"?: string;
+        /** Maximum reported image pixel count in megapixels before decoding. Default: 50. */
+        "image-max-megapixels"?: string;
+        /** Alias for image-max-megapixels. Set only one spelling. */
+        image_max_megapixels?: string;
         /**
          * Run the multimodal projector (mmproj / vision encoder) on the GPU. Accepts
          * 'true'/'on'/'1' or 'false'/'off'/'0'. When unset, the backend is auto-selected

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qvac/llm-llamacpp",
-  "version": "0.60.0",
+  "version": "0.61.0",
   "description": "llama addon for qvac",
   "addon": true,
   "scripts": {
@@ -73,8 +73,8 @@
   "bugs": "https://github.com/tetherto/qvac/issues",
   "homepage": "https://qvac.tether.io",
   "devDependencies": {
-    "@qvac/fabric-android-arm64": "0.21.1",
-    "@qvac/fabric-ios": "0.21.1",
+    "@qvac/fabric-android-arm64": "0.22.0",
+    "@qvac/fabric-ios": "0.22.0",
     "@types/node": "^24.2.1",
     "bare-crypto": "^1.15.3",
     "bare-process": "^4.2.2",
@@ -96,22 +96,22 @@
     "typescript-eslint": "^8.63.0"
   },
   "dependencies": {
-    "@qvac/fabric": "^0.21.1",
-    "@qvac/infer-base": "^0.6.2",
+    "@qvac/fabric": "^0.22.0",
+    "@qvac/infer-base": "^0.8.0",
     "@qvac/logging": "^0.1.0",
     "bare-fs": "^4.5.1",
     "bare-path": "^3.0.0"
   },
   "optionalDependencies": {
-    "@qvac/fabric-linux-x64": "^0.21.1",
-    "@qvac/fabric-linux-arm64": "^0.21.1",
-    "@qvac/fabric-darwin-arm64": "^0.21.1",
-    "@qvac/fabric-darwin-x64": "^0.21.1",
-    "@qvac/fabric-win32-x64": "^0.21.1"
+    "@qvac/fabric-linux-x64": "^0.22.0",
+    "@qvac/fabric-linux-arm64": "^0.22.0",
+    "@qvac/fabric-darwin-arm64": "^0.22.0",
+    "@qvac/fabric-darwin-x64": "^0.22.0",
+    "@qvac/fabric-win32-x64": "^0.22.0"
   },
   "peerDependencies": {
-    "@qvac/fabric-android-arm64": "^0.21.1",
-    "@qvac/fabric-ios": "^0.21.1"
+    "@qvac/fabric-android-arm64": "^0.22.0",
+    "@qvac/fabric-ios": "^0.22.0"
   },
   "peerDependenciesMeta": {
     "@qvac/fabric-android-arm64": {

package/prebuilds/darwin-arm64/qvac__llm-llamacpp.bare.exports

@@ -7,7 +7,7 @@
     ],
     "current_versions": [
       {
-        "version": "0.60"
+        "version": "0.61"
       }
     ],
     "exported_symbols": [
@@ -39,6 +39,8 @@
             "__ZN28qvac_lib_inference_addon_cpp17MultiJobScheduler14tlWorkerOwner_E"
           ],
           "weak": [
+            "__ZZN30qvac_lib_inference_addon_llama6runFitENS_10FitRequestEE8fitMutex",
+            "__ZGVZN30qvac_lib_inference_addon_llama6runFitENS_10FitRequestEE8fitMutex",
             "__ZNSt3__19__any_imp17__unique_typeinfoIN24llama_finetuning_helpers21FinetuneProgressStatsEE4__idE",
             "__ZZN10InitLoader17getLoadingContextERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEE12numInstances",
             "__ZN28qvac_lib_inference_addon_cpp6logger8JsLogger6state_E",
@@ -47,7 +49,6 @@
             "__ZNSt3__119piecewise_constructE",
             "__ZN8picojson12last_error_tIbE1sE",
             "__ZGVN28qvac_lib_inference_addon_cpp11JsInterface10instances_E",
-            "__ZGVZN30qvac_lib_inference_addon_llama9assessFitEP8js_env_sP18js_callback_info_sE8fitMutex",
             "__ZNSt3__19__any_imp17__unique_typeinfoINS_6vectorINS_4pairINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEENS_7variantIJdxEEEEENS7_ISC_EEEEE4__idE",
             "__ZN28qvac_lib_inference_addon_cpp6logger8JsLogger13logger_async_E",
             "__ZGVN28qvac_lib_inference_addon_cpp6logger8JsLogger10log_queue_E",
@@ -71,7 +72,6 @@
             "__ZNSt3__16ranges5__cpo4findE",
             "__ZN30qvac_lib_inference_addon_llama10JsBatchIds11kMintPrefixE",
             "__ZGVN28qvac_lib_inference_addon_cpp11JsInterface13instancesMtx_E",
-            "__ZZN30qvac_lib_inference_addon_llama9assessFitEP8js_env_sP18js_callback_info_sE8fitMutex",
             "__ZNSt3__19__any_imp17__unique_typeinfoIN28qvac_lib_inference_addon_cpp6Output5ErrorEE4__idE",
             "__ZZN30qvac_lib_inference_addon_llama10JsBatchIds4mintEvE4next",
             "__ZNSt3__19__any_imp17__unique_typeinfoINS_6vectorINS_12basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEENS6_IS8_EEEEE4__idE"
@@ -81,7 +81,6 @@
           "global": [
             "__ZNK14MtmdLlmContext20getVisionEncodeTilesEv",
             "__ZN19LlamaBackendsHandleC2EOS_",
-            "_bare_get_module_name_v0",
             "__ZN24llama_finetuning_helpers23optEpochCallbackWrapperEbP16ggml_opt_contextP16ggml_opt_datasetP15ggml_opt_resultxxx",
             "__ZThn24_N10LlamaModelD0Ev",
             "__ZN14MtmdLlmContext8setNPastEi",
@@ -91,7 +90,6 @@
             "__ZN10LlamaModel30discardFinetuneCancelSaveModesERKNSt3__16vectorIyNS0_9allocatorIyEEEE",
             "__ZN30qvac_lib_inference_addon_llama8batching7RequestC1EjO11PrefillPlanji",
             "__ZN30qvac_lib_inference_addon_llama5utils25partialSnapshotFromStreamERNSt3__113basic_istreamIcNS1_11char_traitsIcEEEEyiNS0_15SnapshotStorageERKNS1_12basic_stringIcS4_NS1_9allocatorIcEEEERKNS1_8functionIFvPKhmEEERNS0_21SequenceStateSnapshotE",
-            "__ZNK13ModelMetaData10isU32OneOfEPKcSt16initializer_listIjE",
             "__ZN10LlamaModelD0Ev",
             "__ZN24llama_finetuning_helpers18schedulerLrForStepERKNS_20LoraLrSchedulerStateEx",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler10setRamTierENSt3__110shared_ptrINS0_14SlotStateCacheEEE",
@@ -120,6 +118,7 @@
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler12processBatchEONSt3__16vectorINS0_13SubmitRequestENS2_9allocatorIS4_EEEEy",
             "__ZN14TextLlmContext6getCtxEv",
             "__ZNK10LlamaModel24singleRuntimeStatsLockedEv",
+            "__ZNK13ModelMetaData10isU32OneOfEPKcSt16initializer_listIjE",
             "__ZN14LlamaFinetuner8finetuneERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsEPN22FinetuneTerminalResult5StatsENSt3__18functionIFvRKN24llama_finetuning_helpers21FinetuneProgressStatsEEEE",
             "__ZN30qvac_lib_inference_addon_llama5utils25selectReasoningBudgetTagsERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEES9_RKNS1_6vectorIS7_NS5_IS7_EEEERKNS1_8optionalINS0_13ReasoningTagsEEE",
             "__ZThn32_N10LlamaModelD0Ev",
@@ -179,10 +178,10 @@
             "__ZN14LlamaFinetuner23initializeCheckpointingERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsEP18llama_adapter_loraPN24llama_finetuning_helpers20LoraLrSchedulerStateE",
             "__ZN14TextLlmContext12tokenizeChatERKNSt3__16vectorI15common_chat_msgNS0_9allocatorIS2_EEEERKNS1_I16common_chat_toolNS3_IS8_EEEERNS1_IiNS3_IiEEEEb",
             "__ZN14MtmdLlmContext17removeLastNTokensEi",
+            "__ZN19LlamaBackendsHandleD2Ev",
             "__ZN24llama_finetuning_helpers25findLatestPauseCheckpointERKNSt3__14__fs10filesystem4pathE",
             "__ZN14MtmdLlmContext35refreshCurrentCacheTokensFromMemoryEv",
             "__ZN30qvac_lib_inference_addon_llama5utils17resolveToolChoiceERKNSt3__18optionalINS1_12basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEEEERKNS1_6vectorI16common_chat_toolNS6_ISD_EEEE",
-            "__ZN19LlamaBackendsHandleD2Ev",
             "__ZN30qvac_lib_inference_addon_llama8batching7RequestC1EjONSt3__16vectorIiNS2_9allocatorIiEEEEji",
             "__ZN30qvac_lib_inference_addon_llama8batching14timeDecodeStepEP13llama_contextR11llama_batchRKNSt3__18functionIFiS2_S4_EEERKNS6_IFvS2_EEE",
             "__ZN30qvac_lib_inference_addon_llama8batching19MultiRequestBatcher20completeMediaBarrierEjiRKNSt3__18functionIFvjimEEE",
@@ -208,9 +207,9 @@
             "__ZN14LlamaFinetuner22prepareDatasetFromPathERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsERKNSt3__112basic_stringIcNS4_11char_traitsIcEENS4_9allocatorIcEEEEPKcSE_",
             "__ZN14LlamaFinetuner26validateModelForFinetuningEv",
             "__ZN10LlamaModel13processPromptERKNS_6PromptE",
-            "__ZN30qvac_lib_inference_addon_llama5utils25requireToolChoiceHonouredE23common_chat_tool_choicebbPKc",
             "__ZN14LlamaFinetuner17createLrSchedulerERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsEx",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler16requestCancelAllEv",
+            "__ZN30qvac_lib_inference_addon_llama5utils25requireToolChoiceHonouredE23common_chat_tool_choicebbPKc",
             "__ZN14TextLlmContext22commitCancelledRequestERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE",
             "__ZN12CacheManager15saveBeforeResetEv",
             "__ZNK10LlamaModel23batchRuntimeStatsLockedEv",
@@ -224,11 +223,14 @@
             "__ZNK30qvac_lib_inference_addon_llama8batching20RuntimeStatsSnapshot22prefillTokensPerSecondEv",
             "__ZN10LlamaModel13setInitLoaderENSt3__18optionalIN10InitLoader11LOADER_TYPEEEENS1_I23FinetuneConfigOverridesEE",
             "__ZNK30qvac_lib_inference_addon_llama8batching7Request13tokenToFeedAtEi",
+            "__ZN14MtmdLlmContextC1ER13common_paramsNSt3__110unique_ptrI18common_init_resultNS2_14default_deleteIS4_EEEEy",
             "__ZNK14MtmdLlmContext19residentStateTokensEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching7Request21chunkConsumesAllUnfedEj",
             "__ZN10LlamaModel7processERKNSt3__13anyE",
+            "__ZN14MtmdLlmContextC1ERK13common_paramsRK15LlmModelContextP12mtmd_contextiiy",
             "__ZThn48_N14MtmdLlmContext8onCancelERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE",
             "__ZN14MtmdLlmContext29adoptCheckpointsFromCacheFileERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEy",
+            "__ZN26LlamaLazyInitializeBackend16initializeLockedERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES8_",
             "__ZN10LlamaModel32setFinetuneCancelSavesCheckpointEbRKNSt3__16vectorIyNS0_9allocatorIyEEEE",
             "__ZN13ModelMetaData5parseERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEERK10GGUFShardsbPKc",
             "__ZNK10LlamaModel15cancelInferenceEv",
@@ -272,14 +274,13 @@
             "__ZThn48_N14TextLlmContext32snapshotPreRequestRollbackAnchorEv",
             "__ZN30qvac_lib_inference_addon_llama5utils16isMedPsyBasenameENSt3__117basic_string_viewIcNS1_11char_traitsIcEEEE",
             "__ZN12CacheManager17acceptLoadedStateERNSt3__16vectorIiNS0_9allocatorIiEEEERKNS0_12basic_stringIcNS0_11char_traitsIcEENS2_IcEEEE",
-            "__ZN14MtmdLlmContextC1ER13common_paramsNSt3__110unique_ptrI18common_init_resultNS2_14default_deleteIS4_EEEE",
             "__ZN14TextLlmContext8onCancelERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE",
+            "__ZN14TextLlmContext32snapshotPreRequestRollbackAnchorEv",
             "__ZN14LlamaFinetuner18configureOptimizerERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsEP18llama_adapter_loraRN24llama_finetuning_helpers20LoraLrSchedulerStateEPNS6_23TrainingCheckpointStateEb",
             "__ZThn40_NK10LlamaModel15consumeJobStatsEy",
             "__ZN17backend_selection30getUnknownFinetuneArchitectureEPK13ModelMetaData",
-            "__ZN14TextLlmContext32snapshotPreRequestRollbackAnchorEv",
+            "__ZThn48_N14MtmdLlmContext32snapshotPreRequestRollbackAnchorEv",
             "__ZN14TextLlmContextC1ERK13common_paramsRK15LlmModelContextii",
-            "__ZN14MtmdLlmContextC1ERK13common_paramsRK15LlmModelContextP12mtmd_contextii",
             "__ZN14MtmdLlmContext29clearCacheReconciliationStateEv",
             "__ZN30qvac_lib_inference_addon_llama8batching7Request18isOptAwaitingMediaERKNSt3__18optionalIS1_EE",
             "__ZN30qvac_lib_inference_addon_llama8batching7Request13isOptFinishedERKNSt3__18optionalIS1_EE",
@@ -291,7 +292,6 @@
             "__ZNK14LlamaFinetuner31getCurrentCheckpointStateSharedEv",
             "__ZThn16_N10LlamaModelD1Ev",
             "__ZThn48_N14MtmdLlmContext9loadCacheERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEb",
-            "__ZThn48_N14MtmdLlmContext32snapshotPreRequestRollbackAnchorEv",
             "__ZN10LlamaModelC1EONSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES7_ONS0_13unordered_mapIS6_S6_NS0_4hashIS6_EENS0_8equal_toIS6_EENS4_INS0_4pairIKS6_S6_EEEEEE",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler25finalizeFinishedSequencesEv",
             "__ZN14LlamaFinetuner15saveLoraAdapterEP18llama_adapter_loraRKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsE",
@@ -305,6 +305,7 @@
             "__ZN17backend_selection23getSplitDeviceSelectionEv",
             "__ZN19LlamaBackendsHandleC1EOS_",
             "__ZN12CacheManager11handleCacheER19ParsedPromptPayloadRKNSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEENS2_8functionIFS0_SA_EEESA_b",
+            "__ZN17image_pixel_limit13takeMaxPixelsERNSt3__113unordered_mapINS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES7_NS0_4hashIS7_EENS0_8equal_toIS7_EENS5_INS0_4pairIKS7_S7_EEEEEE",
             "__ZN30qvac_lib_inference_addon_llama5utils21snapshotSequenceStateEP13llama_contextiiRNS0_21SequenceStateSnapshotENS0_15SnapshotStorageENS0_13SnapshotScopeERKNSt3__112basic_stringIcNS7_11char_traitsIcEENS7_9allocatorIcEEEE",
             "__ZNK10LlamaModel16supportsBatchingEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching20RuntimeStatsSnapshot21decodeTokensPerSecondEv",
@@ -335,6 +336,7 @@
             "__ZN14TextLlmContext30capturePreRequestCacheSnapshotEv",
             "__ZN30qvac_lib_inference_addon_llama8batching19MultiRequestBatcher15extractFinishedEv",
             "__ZN14MtmdLlmContext23cancelGenerationCleanupERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE",
+            "__ZN26LlamaLazyInitializeBackend7acquireERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES8_",
             "__ZN14TextLlmContext14requireSamplerEv",
             "__ZN19LlamaBackendsHandleC2ERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES8_",
             "__ZN24llama_finetuning_helpers23checkpointStepDirectoryERKNS_23TrainingCheckpointStateEx",
@@ -377,16 +379,15 @@
             "__ZN14MtmdLlmContext16generateResponseERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler14failSlotLockedEjSt13exception_ptr",
             "__ZNK14LlamaFinetuner17isFinetuneRunningEv",
-            "__ZN14MtmdLlmContextC2ER13common_paramsNSt3__110unique_ptrI18common_init_resultNS2_14default_deleteIS4_EEEE",
             "__ZN30qvac_lib_inference_addon_llama5utils21updateReasoningBufferERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEERNS0_14ReasoningStateE",
             "__ZThn24_N10LlamaModelD1Ev",
             "__ZN17backend_selection23applyAdrenoRestrictionsERNS_20SplitDeviceSelectionERK13ModelMetaDatab",
             "__ZThn48_N14MtmdLlmContext20onGenerationFinishedERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE20GenerationStopReason",
             "__ZN14MtmdLlmContext13resetStopFlagEv",
+            "__ZNK30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler12parkedSeqIdsEv",
             "__ZN30qvac_lib_inference_addon_llama5utils24requireSnapshotDirectoryERKNSt3__112basic_stringIcNS1_11char_traitsIcEENS1_9allocatorIcEEEE",
             "__ZN24llama_finetuning_helpers24pauseCheckpointDirectoryERKNSt3__14__fs10filesystem4pathEx",
             "__Z21computeMediaLoadOrderRKNSt3__16vectorI12PlannedMediaNS_9allocatorIS1_EEEE",
-            "__ZNK30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler12parkedSeqIdsEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler9occupancyEv",
             "__ZThn48_N14MtmdLlmContext24captureHistoryCheckpointEi",
             "__ZThn32_N10LlamaModelD1Ev",
@@ -467,6 +468,7 @@
             "__ZN14TextLlmContext9loadCacheERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEb",
             "__ZN30qvac_lib_inference_addon_llama8batching7Request19isOptPrefillPendingERKNSt3__18optionalIS1_EE",
             "__ZN14MtmdLlmContext21applyGenerationParamsERK16GenerationParams",
+            "__ZN17image_pixel_limit11checkBufferEPKhmy",
             "__ZN10LlamaModel6reloadENSt3__18optionalI23FinetuneConfigOverridesEE",
             "__ZN14LlamaFinetuner18prepareEvalDatasetERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsE",
             "__ZNK13ModelMetaData12tryGetStringEPKc",
@@ -494,13 +496,14 @@
             "__ZN18AsyncWeightsLoader30extractIndividualStreamedFilesEv",
             "__ZN30qvac_lib_inference_addon_llama8batching20computeObservedStatsENSt3__16chrono10time_pointINS2_12steady_clockENS2_8durationIxNS1_5ratioILl1ELl1000000000EEEEEEERKNS0_7RequestENS1_8optionalI20GenerationStopReasonEE",
             "__ZN10LlamaModel15getCommonParamsEv",
+            "__ZN14MtmdLlmContextC2ERK13common_paramsRK15LlmModelContextP12mtmd_contextiiy",
             "__ZN30qvac_lib_inference_addon_llama5utils19getHarmonyCallTokenEP13llama_context",
             "__ZN30qvac_lib_inference_addon_llama5utils21SequenceStateSnapshotD2Ev",
-            "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler5clearEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler15hasValidDriverFEv",
-            "__ZThn48_NK14MtmdLlmContext9saveCacheERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEE",
+            "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler5clearEv",
             "__ZThn24_N10LlamaModel7processERKNSt3__13anyEy",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler29serviceNextMediaSegmentLockedEPNSt3__111unique_lockINS2_5mutexEEE",
+            "__ZThn48_NK14MtmdLlmContext9saveCacheERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEE",
             "__ZN30qvac_lib_inference_addon_llama5utils13isGemma4ModelEPK11llama_model",
             "__ZN10LlamaModel24validateCheckpointBudgetERNS_15ReloadableStateE",
             "__ZN22load_fit_normalization31validateMobileMultiDeviceConfigERKNSt3__113unordered_mapINS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES7_NS0_4hashIS7_EENS0_8equal_toIS7_EENS5_INS0_4pairIKS7_S7_EEEEEE16llama_split_mode",
@@ -559,7 +562,6 @@
             "__ZN30qvac_lib_inference_addon_llama5utils21SequenceStateSnapshotC2EOS1_",
             "__ZN13ModelMetaData28FirstFileFromGgufStreamState7provideER13BorrowablePtrINSt3__115basic_streambufIcNS2_11char_traitsIcEEEEE",
             "__ZN10LlamaModel20isConcurrentEligibleERKNS_6PromptE",
-            "__ZN14MtmdLlmContextC2ERK13common_paramsRK15LlmModelContextP12mtmd_contextii",
             "__ZNK30qvac_lib_inference_addon_llama8batching7Request15isAwaitingMediaEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler15numActiveLockedEv",
             "__ZN10LlamaModelD1Ev",
@@ -570,19 +572,20 @@
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler6cancelEjy",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler17resetRuntimeStatsEv",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler16saveConversationERKNSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEE",
+            "__ZN14MtmdLlmContextC2ER13common_paramsNSt3__110unique_ptrI18common_init_resultNS2_14default_deleteIS4_EEEEy",
             "__ZN14TextLlmContext10resetStateEb",
-            "__ZN24llama_finetuning_helpers15tokenizeDatasetEP13llama_contextRKNSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEE",
             "__ZThn48_N14MtmdLlmContext14preparePrefillERKNSt3__16vectorI15common_chat_msgNS0_9allocatorIS2_EEEERKNS1_I16common_chat_toolNS3_IS8_EEEERKNS1_INS1_IhNS3_IhEEEENS3_ISE_EEEERKNS1_I12PlannedMediaNS3_ISJ_EEEEbb",
-            "__ZThn48_NK14TextLlmContext25getToolDefinitionsDroppedEv",
+            "__ZN24llama_finetuning_helpers15tokenizeDatasetEP13llama_contextRKNSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEE",
             "__ZN17backend_selection17tryMainGpuFromMapERNSt3__113unordered_mapINS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES7_NS0_4hashIS7_EENS0_8equal_toIS7_EENS5_INS0_4pairIKS7_S7_EEEEEE",
             "__ZN14LlamaFinetuner30setPausedCheckpointStateSharedENSt3__110shared_ptrIN24llama_finetuning_helpers23TrainingCheckpointStateEEE",
+            "__ZThn48_NK14TextLlmContext25getToolDefinitionsDroppedEv",
             "__ZN30qvac_lib_inference_addon_llama5utils21SequenceStateSnapshot14seedForTestingENSt3__112basic_stringIcNS2_11char_traitsIcEENS2_9allocatorIcEEEEi",
             "__ZN19LlamaBackendsHandleC1ERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES8_",
             "__ZN10LlamaModel7processERKNSt3__13anyEy",
             "__ZN24llama_finetuning_helpers24resolveAdapterOutputPathERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsE",
-            "__ZN24llama_finetuning_helpers14saveCheckpointEP16ggml_opt_contextRNS_23TrainingCheckpointStateE",
+            "__ZN17image_pixel_limit15readRegularFileERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEE",
             "__ZN12CacheManager39discardActiveCacheIfBackingStoreMissingEv",
-            "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler10workerLoopEv",
+            "__ZN24llama_finetuning_helpers14saveCheckpointEP16ggml_opt_contextRNS_23TrainingCheckpointStateE",
             "__Z34applyGenerationOverridesToSamplingR22common_params_samplingRiRK16GenerationParams",
             "__ZN10LlamaModel17setWeightsForFileERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEONS0_10unique_ptrINS0_15basic_streambufIcS3_EENS0_14default_deleteISB_EEEE",
             "__ZN30qvac_lib_inference_addon_llama5utils30createPrivateSnapshotDirectoryERKNSt3__14__fs10filesystem4pathE",
@@ -590,6 +593,8 @@
             "__ZN14TextLlmContext22configureReasoningTagsERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEES8_RKNS0_8optionalIN30qvac_lib_inference_addon_llama5utils13ReasoningTagsEEE",
             "__ZN12CacheManager15hasTurnsToFlushEv",
             "__ZThn48_N14MtmdLlmContext18adoptResidentStateERKNSt3__16vectorIiNS0_9allocatorIiEEEE",
+            "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler10workerLoopEv",
+            "__ZN17image_pixel_limit9checkFileERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEy",
             "__ZN14MtmdLlmContext27resetToolDefinitionsDroppedEv",
             "__ZNK30qvac_lib_inference_addon_llama8batching19MultiRequestBatcher12chunkSizeForEj",
             "__ZNK14LlamaFinetuner30getPausedCheckpointStateSharedEv",
@@ -597,6 +602,7 @@
             "__ZThn48_N10LlamaModelD1Ev",
             "__ZThn48_N14TextLlmContext17onPrefillCompleteEim",
             "__ZN30qvac_lib_inference_addon_llama8batching24ContinuousBatchScheduler14flushForUnloadEv",
+            "_bare_get_module_name_v0",
             "__ZN14LlamaFinetuner22prepareTrainingDatasetERKN30qvac_lib_inference_addon_llama21LlamaFinetuningParamsE",
             "__ZNK30qvac_lib_inference_addon_llama8batching20RuntimeStatsSnapshot9elapsedMsEv",
             "__ZN14TextLlmContext13resetStopFlagEv",
@@ -640,14 +646,18 @@
             "__ZN14MtmdLlmContext13onSequenceEndERKNSt3__18functionIFvRKNS0_12basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEEEE"
           ],
           "weak": [
+            "__ZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask13onEnvTeardownIN30qvac_lib_inference_addon_llama9FitResultEEEvP22js_deferred_teardown_sPv",
             "__ZZZN28qvac_lib_inference_addon_cpp16OutputCallBackJs26initializeProcessingThreadENSt3__110shared_ptrINS_11OutputQueueEEEENKUlvE_clEvENUlP11uv_handle_sE_8__invokeES7_",
             "__ZN30qvac_lib_inference_addon_llama9assessFitEP8js_env_sP18js_callback_info_s",
             "__ZN30qvac_lib_inference_addon_llama6cancelEP8js_env_sP18js_callback_info_s",
+            "__ZN30qvac_lib_inference_addon_llama15fitResultObjectEP8js_env_sRKNS_9FitResultE",
             "__ZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask13onEnvTeardownINSt3__19monostateEEEvP22js_deferred_teardown_sPv",
             "__ZN30qvac_lib_inference_addon_llama11activeSlotsEP8js_env_sP18js_callback_info_s",
+            "__ZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask13onCloseHandleIN30qvac_lib_inference_addon_llama9FitResultEEEvP11uv_handle_s",
             "__ZN28qvac_lib_inference_addon_cpp6logger8JsLogger13asyncCallbackEP10uv_async_s",
             "__ZN28qvac_lib_inference_addon_cpp11JsInterface6cancelEP8js_env_sP18js_callback_info_s",
             "__ZN28qvac_lib_inference_addon_cpp11JsInterface10activeJobsEP8js_env_sP18js_callback_info_s",
+            "__ZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask10onCompleteIN30qvac_lib_inference_addon_llama9FitResultEEEvP10uv_async_s",
             "__ZZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask5startINSt3__19monostateEEEP10js_value_sP8js_env_sNS1_12CallbackDataIT_E4WorkENSB_7MarshalEENUlP11uv_handle_sE_8__invokeESF_",
             "__ZN30qvac_lib_inference_addon_llama12discardCacheEP8js_env_sP18js_callback_info_s",
             "__ZN30qvac_lib_inference_addon_llama8finetuneEP8js_env_sP18js_callback_info_s",
@@ -657,6 +667,7 @@
             "__ZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask10onCompleteINSt3__19monostateEEEvP10uv_async_s",
             "__ZZN28qvac_lib_inference_addon_cpp6logger8JsLogger22closeAsyncHandleLockedEvENUlP11uv_handle_sE_8__invokeES3_",
             "__ZN30qvac_lib_inference_addon_llama14createInstanceEP8js_env_sP18js_callback_info_s",
+            "__ZZN28qvac_lib_inference_addon_cpp2js11JsAsyncTask5startIN30qvac_lib_inference_addon_llama9FitResultEEEP10js_value_sP8js_env_sNS1_12CallbackDataIT_E4WorkENSB_7MarshalEENUlP11uv_handle_sE_8__invokeESF_",
             "__ZN30qvac_lib_inference_addon_llama6runJobEP8js_env_sP18js_callback_info_s",
             "__ZN30qvac_lib_inference_addon_llama9saveCacheEP8js_env_sP18js_callback_info_s",
             "__ZN28qvac_lib_inference_addon_cpp11JsInterface15destroyInstanceEP8js_env_sP18js_callback_info_s",