Skip to content

Fix compatibility with MLX-Swift-Examples v2.25.7 #59

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix compatibility with MLX-Swift-Examples v2.25.7 #59

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
4 changes: 4 additions & 0 deletions .gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -136,6 +136,10 @@ dmypy.json

# Xcode
*.xcworkspace
xcuserdata/
*.xcuserstate
*.xcuserdatad/
project.pbxproj.orig

# FastVLM models
app/FastVLM/model
36 changes: 27 additions & 9 deletions app/FastVLM/FastVLM.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -213,10 +213,20 @@ private enum Language {
fatalError("one of inputs or inputEmbedding must be non-nil")
}

let mask = createAttentionMask(h: h, cache: cache)
let mask = createAttentionMask(h: h, cache: cache, returnArray: true)
let maskArray: MLXArray? = {
switch mask {
case .array(let array):
return array
case .causal, .none:
return nil
case .arrays(_):
return nil
}
}()

for (i, layer) in layers.enumerated() {
h = layer(h, mask: mask, cache: cache?[i])
h = layer(h, mask: maskArray, cache: cache?[i])
}

return norm(h)
Expand Down Expand Up @@ -361,13 +371,21 @@ public class FastVLMProcessor: UserInputProcessor {
}

public func prepare(prompt: UserInput.Prompt, imageTHW: THW?) -> String {
var messages = prompt.asMessages()
if messages[0]["role"] != "system" {
var messages: [Message]
switch prompt {
case .text(let text):
messages = [["role": "user", "content": text]]
case .messages(let msgs):
messages = msgs
case .chat(let chatMsgs):
messages = chatMsgs.map { ["role": $0.role.rawValue, "content": $0.content] }
}
if (messages[0]["role"] as? String) != "system" {
messages.insert(["role": "system", "content": "You are a helpful assistant."], at: 0)
}

let lastIndex = messages.count - 1
var lastMessage = messages[lastIndex]["content"] ?? ""
var lastMessage: String = (messages[lastIndex]["content"] as? String) ?? ""

// processing_llava.py
if let imageTHW {
Expand All @@ -382,8 +400,8 @@ public class FastVLMProcessor: UserInputProcessor {
numImageTokens -= 1
}

lastMessage += Array(repeating: config.imageToken, count: numImageTokens)
.joined()
let imageTokens = String(repeating: config.imageToken, count: numImageTokens)
lastMessage += imageTokens
}

messages[lastIndex]["content"] = lastMessage
Expand Down Expand Up @@ -411,7 +429,7 @@ public class FastVLMProcessor: UserInputProcessor {

let (pixels, thw) = try preprocess(
image: input.images[0].asCIImage(), processing: input.processing)
let image = LMInput.ProcessedImage(pixels: pixels, imageGridThw: [thw])
let image = LMInput.ProcessedImage(pixels: pixels, frames: [thw])

let prompt = prepare(prompt: input.prompt, imageTHW: thw)
let promptTokens = tokenizer.encode(text: prompt)
Expand Down Expand Up @@ -537,7 +555,7 @@ public class FastVLM: Module, VLMModel, KVCacheDimensionProvider {
public func prepare(_ input: LMInput, cache: [any KVCache], windowSize: Int?) throws
-> PrepareResult
{
let gridThw = input.image?.imageGridThw
let gridThw = input.image?.frames

let dtype = DType.float32
let pixels = input.image?.pixels.asType(dtype)
Expand Down