Hello! I'm following the Foundation Models adapter training guide (https://developer.apple.com/apple-intelligence/foundation-models-adapter/) on my NVIDIA DGX Spark box. I'm able to train on my own data but the example notebook fails when I try to export the artifact as an fmadapter. I get the following error for the code block I'm trying to run. I haven't touched any of the code in the export folder. I tried exporting it on my Mac too and got the same error as well (given below). Would appreciate some more clarity around this. Thank you. Code Block: from export.export_fmadapter import Metadata, export_fmadapter metadata = Metadata( author="3P developer", description="An adapter that writes play scripts.", ) export_fmadapter( output_dir="./", adapter_name="myPlaywritingAdapter", metadata=metadata, checkpoint="adapter-final.pt", draft_checkpoint="draft-model-final.pt", ) Error: --------------------------------------------------------------------------- ModuleNotFoundError Traceback (most recent call last) Cell In[10], line 1 ----> 1 from export.export_fmadapter import Metadata, export_fmadapter 3 metadata = Metadata( 4 author="3P developer", 5 description="An adapter that writes play scripts.", 6 ) 8 export_fmadapter( 9 output_dir="./", 10 adapter_name="myPlaywritingAdapter", (...) 13 draft_checkpoint="draft-model-final.pt", 14 ) File /workspace/export/export_fmadapter.py:11 8 from typing import Any 10 from .constants import BASE_SIGNATURE, MIL_PATH ---> 11 from .export_utils import AdapterConverter, AdapterSpec, DraftModelConverter, camelize 13 logger = logging.getLogger(__name__) 16 class MetadataKeys(enum.StrEnum): File /workspace/export/export_utils.py:15 13 import torch 14 import yaml ---> 15 from coremltools.libmilstoragepython import _BlobStorageWriter as BlobWriter 16 from coremltools.models.neural_network.quantization_utils import _get_kmeans_lookup_table_and_weight 17 from coremltools.optimize._utils import LutParams ModuleNotFoundError: No module named 'coremltools.libmilstoragepython'

Is it possible to train a model using CreateML to infer a relevance numeric score of a news article based on similar trained data, something like a sentiment score ? I created a Text Classifier that assigns a category label which works perfect but I would like a solution that calculates a numeric value, not a label.

Hello Apple Developer Community, I'm investigating Core ML model loading behavior and noticed that even when the compiled model path remains unchanged after an APP update, the first run still triggers an "uncached load" process. This seems to impact user experience with unnecessary delays. Question: Does Core ML provide any public API to check whether a compiled model (from a specific .mlmodelc path) is already cached in the system? If such API exists, we'd like to use it for pre-loading decision logic - only perform background pre-load when the model isn't cached. Has anyone encountered similar scenarios or found official solutions? Any insights would be greatly appreciated!

the specific context is that i would like to build an agent that monitors my phone call (with a customer support for example), and simiply identify whether or not im still put on hold, and notify me when im not. currently after reading the doc, i dont think its possible yet, but im so annoyed by the customer support calls that im willing to go the distance and see if theres any way.

Bear with me, please. Please make sure a highly skilled technical person reads and understands this. I want to describe my vision for (AI/Algorithmically) Optimised Operating Systems. To explain it properly, I will describe the process to build it (pseudo). Required Knowledge (no particular order): Processor Logic Circuits, LLM models, LLM tool usage, Python OO coding, Procedural vs OO, NLP fuzzy matching, benchmarking, canvas/artefacts/dynamic HTML interfaces, concepts of how AI models are vastly compressed and miniaturised forms of full data, Algorithmic vs AI. First, take all OO Python code (example) on GitHub (example), then separate each function from each object into its own procedure (procedural logic) by making a logical procedural list of actions to perform only that function based on its entire dependency chain (i.e. all other objects it relies on). Relate all compiled functions using (for example) fuzzy matching on the name, or AI-based functional profiling to get multiple instances of each function. Starting with the most used function, test each one against the others that perform the same task for bugs and completeness. Determine the fastest, most optimal version of that function (and every function). Add a single instance of each most optimal function to the centralised tool codebase, which will later be utilised by the language models. This ensures we rely only on the most optimised function for each and every use case — with every program using one shared instance of that function instead of compiling it separately. In this way, we optimise all existing (Python OO) code based on maximum availability of certain resources. Use the optimised code to teach the AI how to map each user task directly to optimised code (i.e. each user-facing function in unoptimised program code to its optimised version). Later, teach the codebase how to optimise these functions directly to the logic circuits on the processor — as an Operating System — using essentially the same process. Eventually, use the most optimised form for tasks via a dynamic canvas/artefacts UI interface generated by the Start Menu using the language model. Allow custom interfaces to emerge dynamically from the available functions based on language understanding. Grow and optimise the algorithmic codebase until the entire operating system — with versioning — evolves based on the AI-generated user interfaces (compressing them into new versions of the tools codebase). Please don’t give up until you understand this concept completely and nothing in the description is beyond your full comprehension. It’s a highly significant step forward.

Recently, I'm trying to deploy some third-party LLM to Apple devices. The methodoloy is similar to https://github.com/Anemll/Anemll. The biggest issue I'm having now is the runtime memory usage. When there are multiple functions in a model (mlpackage or mlmodelc), the runtime memory usage for weights is somehow duplicated when I load all of them. Here's the detail: I created my multifunction mlpackage following https://apple.github.io/coremltools/docs-guides/source/multifunction-models.html I loaded each of the functions using the generated swift class: let config = MLModelConfiguration() config.computeUnits = MLComputeUnits.cpuAndNeuralEngine config.functionName = "infer_512"; let ffn1_infer_512 = try! mimo_FFN_PF_lut4_chunk_01of02(configuration: config) config.functionName = "infer_1024"; let ffn1_infer_1024 = try! mimo_FFN_PF_lut4_chunk_01of02(configuration: config) config.functionName = "infer_2048"; let ffn1_infer_2048 = try! mimo_FFN_PF_lut4_chunk_01of02(configuration: config) I observed that RAM usage increases linearly as I load each of the functions. Using instruments, I see that there are multiple HWX files generated and loaded, each of which contains all the weight data. My understanding of what's happening here: The CoreML framework did some MIL->MIL preprocessing before further compilation, which includes separating CPU workload from ANE workload. The ANE part of each function is moved into a separate MIL file then compile separately into a HWX file each. The problem is that the weight data of these HWX files are duplicated. Since that the weight data of LLMs is huge, it will cause out-of-memory issue on mobile devices. The improvement I'm hoping from Apple: I hope we can try to merge the processed MIL files back into one before calling ANECCompile(), so that the weights can be merged. I don't have control over that in user space and I'm not sure if that is feasible. So I'm asking for help here. Thanks.

Hello everyone, I have a visual convolutional model and a video that has been decoded into many frames. When I perform inference on each frame in a loop, the speed is a bit slow. So, I started 4 threads, each running inference simultaneously, but I found that the speed is the same as serial inference, every single forward inference is slower. I used the mactop tool to check the GPU utilization, and it was only around 20%. Is this normal? How can I accelerate it?

Hello. I am willing to hire game developer for cards game called baloot. My question is Can the developer implement an AI when the computer is playing and the computer on the same time the conputer improves his rises level without any interaction? 🌹

Hi all. My adapter model just won't invoke my tool. The problem I am having is covered in an older post: https://developer.apple.com/forums/thread/794839?answerId=852262022#852262022 Sadly the thread dies there and no resolution is seen in that thread. It's worth noting that I have developed an AI chatbot built around LanguageModelSession to which I feed the exact same system prompt that I feed to my training set (pasted further in this post). The AI chatbot works perfectly, the tool is invoked when needed. I am training the adapter model because the base model whilst capable doesn't produce the quality I'm looking for. So here's the template of an item in my training set: [ { 'role': 'system', 'content': systemPrompt, 'tools': [TOOL_DEFINITION] }, { 'role': 'user', 'content': entry['prompt'] }, { 'role': 'assistant', 'content': entry['code'] } ] where TOOL_DEFINITION = { 'type': 'function', 'function': { 'name': 'WriteUbersichtWidgetToFileSystem', 'description': 'Writes an Übersicht Widget to the file system. Call this tool as the last step in processing a prompt that generates a widget.', 'parameters': { 'type': 'object', 'properties': { 'jsxContent': { 'type': 'string', 'description': 'Complete JSX code for an Übersicht widget. This should include all required exports: command, refreshFrequency, render, and className. The JSX should be a complete, valid Übersicht widget file.' } }, 'required': ['jsxContent'] } } ... and systemPrompt = A conversation between a user and a helpful assistant. You are an Übersicht widget designer. Create Übersicht widgets when requested by the user. IMPORTANT: You have access to a tool called WriteUbersichtWidgetToFileSystem. When asked to create a widget, you MUST call this tool. ### Tool Usage: Call WriteUbersichtWidgetToFileSystem with complete JSX code that implements the Übersicht Widget API. Generate custom JSX based on the user's specific request - do not copy the example below. ### Übersicht Widget API (REQUIRED): Every Übersicht widget MUST export these 4 items: - export const command: The bash command to execute (string) - export const refreshFrequency: Refresh rate in milliseconds (number) - export const render: React component function that receives {output} prop (function) - export const className: CSS positioning for absolute placement (string) Example format (customize for each request): WriteUbersichtWidgetToFileSystem({jsxContent: `export const command = "echo hello"; export const refreshFrequency = 1000; export const render = ({output}) => { return <div>{output}</div>; }; export const className = "top: 20px; left: 20px;"`}) ### Rules: - The terms "ubersicht widget", "widget", "a widget", "the widget" must all be interpreted as "Übersicht widget" - Generate complete, valid JSX code that follows the Übersicht widget API - When you generate a widget, don't just show JSON or code - you MUST call the WriteUbersichtWidgetToFileSystem tool - Report the results to the user after calling the tool ### Examples: - "Generate a Übersicht widget" → Use WriteUbersichtWidgetToFileSystem tool - "Can you add a widget that shows the time" → Use WriteUbersichtWidgetToFileSystem tool - "Create a widget with a button" → Use WriteUbersichtWidgetToFileSystem tool When the script that I use to compose the full training set is executed, entry['prompt'] and entry['code'] contain the prompt and the resulting JSX code for one of the examples I'm feeding to the training session. This is repeated for about 60 such examples that I have in my sample data collection. Thanks for any help. Michael

I've created a "Transfer Learning BERT Embeddings" model with the default "Latin" language family and "Automatic" Language setting. This model performs exceptionally well against the test data set and functions as expected when I preview it in Create ML. However, when I add it to the Xcode project of the application to which I am deploying it, I am getting runtime errors that suggest it can't find the embedding resources: Failed to locate assets for 'mul_Latn' - '5C45D94E-BAB4-4927-94B6-8B5745C46289' embedding model Note, I am adding the model to the app project the same way that I added an earlier "Maximum Entropy" model. That model had no runtime issues. So it seems there is an issue getting hold of the embeddings at runtime. For now, "runtime" means in the Simulator. I intend to deploy my application to iOS devices once GM 26 is released (the app also uses AFM). I'm developing on Tahoe 26 beta, running on iOS 26 beta, using Xcode 26 beta. Is this a known/expected issue? Are the embeddings expected to be a resource in the model? Is there a workaround? I did try opening the model in Xcode and saving it as an mlpackage, then adding that to my app project, but that also didn't resolve the issue.

JAX Metal shows 55x slower random number generation compared to NVIDIA CUDA on equivalent workloads. This makes Monte Carlo simulations and scientific computing impractical on Apple Silicon. Performance Comparison NVIDIA GPU: 0.475s for 12.6M random elements M1 Max Metal: 26.3s for same workload Performance gap: 55x slower Environment Apple M1 Max, 64GB RAM, macOS Sequoia Version 15.6.1 JAX 0.4.34, jax-metal latest Backend: Metal Reproduction Code import time import jax import jax.numpy as jnp from jax import random key = random.PRNGKey(42) start_time = time.time() random_array = random.normal(key, (50000, 252)) duration = time.time() - start_time print(f"Duration: {duration:.3f}s")

Hi, I am modifying the sample camera app that is here: https://developer.apple.com/tutorials/sample-apps/capturingphotos-camerapreview ... In the processPreviewImages, I am using the Vision APIs to generate a segmentation mask for a person/object, then compositing that person onto a different background (with some other filtering). The filtering and compositing is done via CoreImage. At the end, I convert the CIImage to a CGImage then to a SwiftUI Image. When I run it on my iPhone, it works fine, and has not crashed. When I run it on the iPhone with the debugger, it crashes within a few seconds with: EXC_BAD_ACCESS in libRPAC.dylib`std::__1::__hash_table<std::__1::__hash_value_type<long, qos_info_t>, std::__1::__unordered_map_hasher<long, std::__1::__hash_value_type<long, qos_info_t>, std::__1::hash, std::__1::equal_to, true>, std::__1::__unordered_map_equal<long, std::__1::__hash_value_type<long, qos_info_t>, std::__1::equal_to, std::__1::hash, true>, std::__1::allocator<std::__1::__hash_value_type<long, qos_info_t>>>::__emplace_unique_key_args<long, std::__1::piecewise_construct_t const&, std::__1::tuple<long const&>, std::__1::tuple<>>: It had previously been working fine with the debugger, so I'm not sure what has changed. Is there a difference in how the Vision APIs are executed if the debugger is attached vs. not?

Hi, I am a new IOS developer, trying to learn to integrate the Apple Foundation Model. my set up is: Mac M1 Pro MacOS 26 Beta Version 26.0 beta 3 Apple Intelligence &amp; Siri --&gt; On here is the code, func generate() { Task { isGenerating = true output = "⏳ Thinking..." do { let session = LanguageModelSession( instructions: """ Extract time from a message. Example Q: Golfing at 6PM A: 6PM """) let response = try await session.respond(to: "Go to gym at 7PM") output = response.content } catch { output = "❌ Error:, \(error)" print(output) } isGenerating = false } and I get these errors guardrailViolation(FoundationModels.LanguageModelSession.GenerationError.Context(debugDescription: "Prompt may contain sensitive or unsafe content", underlyingErrors: [Asset com.apple.gm.safety_embedding_deny.all not found in Model Catalog])) Can you help me get through this?

Hi all, I noticed on Friday that on the new Beta 5 using FoundationModels on a simulator LanguageModelSession.respond() neither resolves nor throws most of the time. The SwiftUI test app below was working perfectly in Xcode 16 Beta 4 and iOS 26 Beta 4 (simulator). import SwiftUI import FoundationModels struct ContentView: View { var body: some View { VStack { Image(systemName: "globe") .imageScale(.large) .foregroundStyle(.tint) Text("Hello, world!") } .padding() .onAppear { Task { do { let session = LanguageModelSession() let response = try await session.respond(to: "are cats better than dogs ???") print(response.content) } catch { print("error") } } } } } After updating to Xcode 16 Beta 5 and iOS 26 Beta 5 (simulator), the code now often hangs. Occasionally it will work if I toggle Apple Intelligence on and off in Settings, but it’s unreliable.

We are really excited to have introduced the Foundation Models framework in WWDC25. When using the framework, you might have feedback about how it can better fit your use cases. Starting in macOS/iOS 26 Beta 4, the best way to provide feedback is to use #Playground in Xcode. To do so: In Xcode, create a playground using #Playground. Fore more information, see Running code snippets using the playground macro. Reproduce the issue by setting up a session and generating a response with your prompt. In the canvas on the right, click the thumbs-up icon to the right of the response. Follow the instructions on the pop-up window and submit your feedback by clicking Share with Apple. Another way to provide your feedback is to file a feedback report with relevant details. Specific to the Foundation Models framework, it’s super important to add the following information in your report: Language model feedback This feedback contains the session transcript, including the instructions, the prompts, the responses, etc. Without that, we can’t reason the model’s behavior, and hence can hardly take any action. Use logFeedbackAttachment(sentiment:issues:desiredOutput: ) to retrieve the feedback data of your current model session, as shown in the usage example, write the data into a file, and then attach the file to your feedback report. If you believe what you’d report is related to the system configuration, please capture a sysdiagnose and attach it to your feedback report as well. The framework is still new. Your actionable feedback helps us evolve the framework quickly, and we appreciate that. Thanks, The Foundation Models framework team

When using the imagePlaygroundSheet modifier in SwiftUI, the system presets an image playground in a fixed size. Especially on macOS, this modal is rather small and doesn't utilize the available space efficiently. Is there a way to make the modal bigger, or allow the user to resize the dialog? I tried presentationDetents, but this would need to be applied to the content of the sheet, which is provided by the system... I guess this question applies to other system-provided sheets like the photo picker as well.

Hi all, I am interested in unlocking unique applications with the new foundational models. I have a few questions regarding the availability of the following features: Image Input: The update in June 2025 mentions "image" 44 times (https://machinelearning.apple.com/research/apple-foundation-models-2025-updates) - however I can't seem to find any information about having images as the input/prompt for the foundational models. When will this be available? I understand that there are existing Vision ML APIs, but I want image input into a multimodal on-device LLM (VLM) instead for features like "Which player is holding the ball in the image", etc (image understanding) Cloud Foundational Model - when will this be available? Thanks! Clement :)

Hi all, When executing an HLO program using the JAX metal PJRT plugin, the program fails due to an unsupported data type returned by the rng_bit_generator operation. The generated HLO includes: %output_state, %output = "mhlo.rng_bit_generator"(%1) <{rng_algorithm = #mhlo.rng_algorithm<PHILOX>}> : (tensor<3xi64>) -> (tensor<3xi64>, tensor<3xui32>) The error message indicates that: Metal only supports MPSDataTypeFloat16, MPSDataTypeBFloat16, MPSDataTypeFloat32, MPSDataTypeInt32, and MPSDataTypeInt64. The use of ui32 seems to be incompatible with Metal’s allowed types. I’m trying to understand if the ui32 output is the problem or maybe the use of rng_bit_generator is wrong. Could you clarify if there is a workaround or planned support for ui32 output in this context? Alternatively, guidance on configuring rng_bit_generator for compatibility with Metal’s supported types would be greatly appreciated.

During testing the “Bringing advanced speech-to-text capabilities to your app” sample app demonstrating the use of iOS 26 SpeechAnalyzer, I noticed that the language model for the English locale was presumably already downloaded. Upon checking the documentation of AssetInventory, I found out that indeed, the language model can be preinstalled on the system. Can someone from the dev team share more info about what assets are preinstalled by the system? For example, can we safely assume that the English language model will almost certainly be already preinstalled by the OS if the phone has the English locale?

I'm using Numbers to build a spreadsheet that I'm exporting as a CSV. I then import this file into Create ML to train a word tagger model. Everything has been working fine for all the models I've trained so far, but now I'm coming across a use case that has been breaking the import process: commas within the training data. This is a case that none of Apple's examples show. My project takes Navajo text that has been tokenized by syllables and labels the parts-of-speech. Case that works... Raw text: Naaltsoos yídéeshtah. Tokens column: Naal,tsoos, ,yí,déesh,tah,. Labels column: NObj,NObj,Space,Verb,Verb,VStem,Punct Case that breaks... Raw text: óola, béésh łigaii, tłʼoh naadą́ą́ʼ, wáin, akʼah, dóó á,shįįh Tokens column with tokenized text (commas quoted): óo,la,",", ,béésh, ,łi,gaii,",", ,tłʼoh, ,naa,dą́ą́ʼ,",", ,wáin,",", ,a,kʼah,",", ,dóó, ,á,shįįh (Create ML reports mismatched columns) Tokens column with tokenized text (commas escaped): óo,la,\,, ,béésh, ,łi,gaii,\,, ,tłʼoh, ,naa,dą́ą́ʼ,\,, ,wáin,\,, ,a,kʼah,\,, ,dóó, ,á,shįįh (Create ML reports mismatched columns) Tokens column with tokenized text (commas escape-quoted): óo,la,\",\", ,béésh, ,łi,gaii,\",\", ,tłʼoh, ,naa,dą́ą́ʼ,\",\", ,wáin,\",\", ,a,kʼah,\",\", ,dóó, ,á,shįįh (record not detected by Create ML) Tokens column with tokenized text (commas escape-quoted): óo,la,"","", ,béésh, ,łi,gaii,"","", ,tłʼoh, ,naa,dą́ą́ʼ,"","", ,wáin,"","", ,a,kʼah,"","", ,dóó, ,á,shįįh (Create ML reports mismatched columns) Labels column: NSub,NSub,Punct,Space,NSub,Space,NSub,NSub,Punct,Space,NSub,Space,NSub,NSub,Punct,Space,NSub,Punct,Space,NSub,NSub,Punct,Space,Conj,Space,NSub,NSub Sample From Spreadsheet Solution Needed It's simple enough to escape commas within CSV files, but the format needed by Create ML essentially combines entire CSV records into single columns, so I'm ending up needing a CSV record that contains a mixture of commas to use for parsing and ones to use as character literals. That's where this gets complicated. For this particular use case (which seems like it would frequently arise when training a word tagger model), how should I properly escape a comma literal?