Skip to content

Open source AI coding agent. Designed for large projects and real world tasks.

Fix tool calls not supported for XML models. Fixes #238

Dane Schneider64b7e9db4d10
diff --git a/app/server/model/model_request.go b/app/server/model/model_request.go
index 7f0b4ee8..c483d38c 100644
--- a/app/server/model/model_request.go
+++ b/app/server/model/model_request.go
@@ -8,6 +8,7 @@ import (
"plandex-server/hooks"
"plandex-server/types"
shared "plandex-shared"
+ "strings"
"time"
"github.com/sashabaranov/go-openai"
@@ -67,7 +68,7 @@ func ModelRequest(
}
messages = FilterEmptyMessages(messages)
-
+ messages = CheckSingleSystemMessage(modelConfig, messages)
inputTokensEstimate := GetMessagesTokenEstimate(messages...) + TokensPerRequest
config := modelConfig.GetRoleForInputTokens(inputTokensEstimate)
@@ -80,12 +81,17 @@ func ModelRequest(
log.Printf("Model config - role: %s, model: %s, max output tokens: %d\n", modelConfig.Role, modelConfig.BaseModelConfig.ModelName, modelConfig.BaseModelConfig.MaxOutputTokens)
+ expectedOutputTokens := modelConfig.BaseModelConfig.MaxOutputTokens - inputTokensEstimate
+ if params.EstimatedOutputTokens != 0 {
+ expectedOutputTokens = params.EstimatedOutputTokens
+ }
+
_, apiErr := hooks.ExecHook(hooks.WillSendModelRequest, hooks.HookParams{
Auth: auth,
Plan: plan,
WillSendModelRequestParams: &hooks.WillSendModelRequestParams{
InputTokens: inputTokensEstimate,
- OutputTokens: modelConfig.BaseModelConfig.MaxOutputTokens - inputTokensEstimate,
+ OutputTokens: expectedOutputTokens,
ModelName: modelConfig.BaseModelConfig.ModelName,
},
})
@@ -101,13 +107,40 @@ func ModelRequest(
reqStarted := time.Now()
req := types.ExtendedChatCompletionRequest{
- Model: modelConfig.BaseModelConfig.ModelName,
- Messages: messages,
- Temperature: modelConfig.Temperature,
- TopP: modelConfig.TopP,
- Stop: stop,
- Tools: tools,
- ToolChoice: toolChoice,
+ Model: modelConfig.BaseModelConfig.ModelName,
+ Messages: messages,
+ }
+
+ if !modelConfig.BaseModelConfig.RoleParamsDisabled {
+ req.Temperature = modelConfig.Temperature
+ req.TopP = modelConfig.TopP
+ }
+
+ if len(tools) > 0 {
+ req.Tools = tools
+ }
+
+ if toolChoice != nil {
+ req.ToolChoice = toolChoice
+ }
+
+ onStream := params.OnStream
+ if modelConfig.BaseModelConfig.StopDisabled {
+ if len(stop) > 0 {
+ onStream = func(chunk string, buffer string) (shouldStop bool) {
+ for _, stopSequence := range stop {
+ if strings.Contains(buffer, stopSequence) {
+ return true
+ }
+ }
+ if params.OnStream != nil {
+ return params.OnStream(chunk, buffer)
+ }
+ return false
+ }
+ }
+ } else {
+ req.Stop = stop
}
if prediction != "" {
@@ -117,12 +150,26 @@ func ModelRequest(
}
}
- res, err := CreateChatCompletionWithInternalStream(clients, modelConfig, ctx, req, params.OnStream, reqStarted)
+ res, err := CreateChatCompletionWithInternalStream(clients, modelConfig, ctx, req, onStream, reqStarted)
if err != nil {
return nil, err
}
+ if modelConfig.BaseModelConfig.StopDisabled && len(stop) > 0 {
+ earliest := len(res.Content)
+ found := false
+ for _, s := range stop {
+ if i := strings.Index(res.Content, s); i != -1 && i < earliest {
+ earliest = i
+ found = true
+ }
+ }
+ if found {
+ res.Content = res.Content[:earliest]
+ }
+ }
+
if params.AfterReq != nil {
params.AfterReq()
}
@@ -204,3 +251,15 @@ func FilterEmptyMessages(messages []types.ExtendedChatMessage) []types.ExtendedC
}
return filteredMessages
}
+
+func CheckSingleSystemMessage(modelConfig *shared.ModelRoleConfig, messages []types.ExtendedChatMessage) []types.ExtendedChatMessage {
+ if len(messages) == 1 && modelConfig.BaseModelConfig.SingleMessageNoSystemPrompt {
+ if messages[0].Role == openai.ChatMessageRoleSystem {
+ msg := messages[0]
+ msg.Role = openai.ChatMessageRoleUser
+ return []types.ExtendedChatMessage{msg}
+ }
+ }
+
+ return messages
+}