phoenix-oss/llama-stack
1
0
Fork 0
forked from phoenix-oss/llama-stack-mirror
Code Pull requests Releases Packages 2 Activity Actions

LocalInferenceImpl update for LS013 (#1242)

Browse source 
# What does this PR do?
[Provide a short summary of what this PR does and why. Link to relevant
issues if applicable.]

[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])

## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]

[//]: # (## Documentation)
This commit is contained in:
Jeff Tang 2025-02-25 09:58:34 -08:00 • committed by GitHub
parent dc3c881ffe
commit 73a0c7a0e7
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 42 additions and 39 deletions
Download patch file Download diff file Expand all files Collapse all files

23
llama_stack/providers/inline/ios/inference/LocalInferenceImpl/Parsing.swift
View file

@ -38,10 +38,10 @@ func encodeMessage(message: Components.Schemas.Message) -> String {
switch (message) {
case .assistant(let m):
if (m.tool_calls.count > 0) {
if (m.tool_calls?.count ?? 0 > 0) {
prompt += "<|python_tag|>"
}
default:
default:0
break
}
@ -91,7 +91,7 @@ func encodeMessage(message: Components.Schemas.Message) -> String {
// for t in m.tool_calls {
// _processContent(t.)
//}
eom = m.stop_reason == Components.Schemas.StopReason.end_of_message
eom = m.stop_reason == Components.Schemas.CompletionMessage.stop_reasonPayload.end_of_message
case .system(_):
break
case .tool(_):
@ -124,8 +124,9 @@ func prepareMessages(request: Components.Schemas.ChatCompletionRequest) throws -
sysContent += try defaultTemplate.render()
messages.append(.system(Components.Schemas.SystemMessage(
content: .case1(sysContent),
role: .system))
role: .system,
content: .case1(sysContent)
))
)
if request.tools?.isEmpty == false {
@ -134,8 +135,8 @@ func prepareMessages(request: Components.Schemas.ChatCompletionRequest) throws -
let toolTemplate = try toolGen.gen(customTools: request.tools!)
let tools = try toolTemplate.render()
messages.append(.user(Components.Schemas.UserMessage(
content: .case1(tools),
role: .user)
role: .user,
content: .case1(tools))
))
}
@ -193,9 +194,9 @@ public func maybeExtractCustomToolCalls(input: String) -> [Components.Schemas.To
result.append(
Components.Schemas.ToolCall(
arguments: .init(additionalProperties: props),
call_id: UUID().uuidString,
tool_name: .case2(name) // custom_tool
tool_name: .case2(name), // custom_tool
arguments: .init(additionalProperties: props)
)
)
}
@ -206,7 +207,7 @@ public func maybeExtractCustomToolCalls(input: String) -> [Components.Schemas.To
}
}
func decodeAssistantMessage(tokens: String, stopReason: Components.Schemas.StopReason) -> Components.Schemas.CompletionMessage {
func decodeAssistantMessage(tokens: String, stopReason: Components.Schemas.CompletionMessage.stop_reasonPayload) -> Components.Schemas.CompletionMessage {
var content = tokens
let roles = ["user", "system", "assistant"]
@ -229,8 +230,8 @@ func decodeAssistantMessage(tokens: String, stopReason: Components.Schemas.StopR
}
return Components.Schemas.CompletionMessage(
content: .case1(content),
role: .assistant,
content: .case1(content),
stop_reason: stopReason,
tool_calls: maybeExtractCustomToolCalls(input: content)
)