| Components | All | New | MacOS | Windows | Linux | iOS | ||||
| Examples | Mac & Win | Server | Client | Guides | Statistic | FMM | Blog | Deprecated | Old | |
Component: Llama
Use llama.cpp in FileMaker to run LLM AI models locally to enjoy some artifical intelligence.
| Version | macOS | Windows | Linux | Server | iOS SDK |
| 16.1 / 16.2 | ✅ Yes | ✅ Yes | ✅ Yes | ✅ Yes | ✅ Yes |
| Item | Details |
|---|---|
| Llama.Backends Queries the backends available. | All 16.1 |
| Llama.Chat Sends a chat messages to the model. | All 16.1 |
| Llama.ChatBuiltinTemplates Queres the list of built-in template names. | All 16.1 |
| Llama.ClearTranscript Clears the current transcript. | All 16.1 |
| Llama.CloseContext Closes the context. | All 16.1 |
| Llama.CloseModel Closes the current model. | All 16.1 |
| Llama.CloseSampler Closes the current sampler. | All 16.1 |
| Llama.ContextAvailable Queries context memory available. | All 16.1 |
| Llama.ContextUsed Queries context memory used. | All 16.1 |
| Llama.Detokenize Convert the provided tokens into text (inverse of llama_tokenize()). | All 16.1 |
| Llama.InitSampler Initializes the sampler chain. | All 16.1 |
| Llama.List Lists all the current llama identifiers. | All 16.1 |
| Llama.LoadBackend Loads a backend for Llama from the given location. | All 16.2 |
| Llama.LoadLibrary Loads the library. | All 16.1 |
| Llama.LoadModel Load the model from a file. | All 16.1 |
| Llama.Loaded Whether the library is loaded. | All 16.1 |
| Llama.ModelChatTemplate Queries the chat template text for the model. | All 16.1 |
| Llama.ModelMetadata Queries the metadata for a model. | All 16.1 |
| Llama.ModelSize Queries the model size. | All 16.1 |
| Llama.New Creates a new session. | All 16.1 |
| Llama.Query Runs a text through the model. | All 16.1 |
| Llama.Release Releases the session. | All 16.1 |
| Llama.ReleaseAll Releases all sessions. | All 16.1 |
| Llama.StartContext Initializes a context. | All 16.1 |
| Llama.Tokenize Convert the provided text into tokens. | All 16.1 |
| Llama.Transcript Queries the transcript. | All 16.1 |
| Llama.TranscriptJSON Queries the transcript as JSON. | All 16.1 |
27 functions shown.
Release notes
- Version 16.3
- Fixed memory leaks in Llama.Tokenize and Llama.Detokenize functions.
- Improved Llama.Chat and Llama.Query to yield time after 1000 tokens.
- Version 16.2
- Added Llama.LoadBackend function.
- Fixed an issue with Llama.Chat when the chat template couldn't be applied causing an exception.
- Version 16.1
- Added Llama functions to load local large language models to query them.
Blog Entries
- MBS FileMaker Plugin 16.2 News
- AI Development with the MBS FileMaker Plugin
- Neues MBS Plugin 16.1 für Claris FileMaker
- MBS FileMaker Plugin 16.1 for Claris FileMaker - More than 7800 functions in one plugin
- Use Llama.cpp in FileMaker
- MBS FileMaker Plugin, version 16.1pr3
Example Databases
MarkDown version: component_Llama.md