xinyin025/one-api
1
0
Fork 0
mirror of https://github.com/songquanpeng/one-api.git synced 2025-12-25 01:05:56 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

feat: refactor AwsClaude to Aws to support both llama3 and claude (#1601)

Browse Source 
* feat: refactor AwsClaude to Aws to support both llama3 and claude

* fix: aws llama3 ratio
This commit is contained in:
Qiying Wang
2024-07-06 13:19:41 +08:00
committed by GitHub
parent e090e76c86
commit 720fe2dfeb
18 changed files with 598 additions and 91 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
relay/controller/audio.go
View File

@@ -7,6 +7,10 @@ import (
"encoding/json"
"errors"
"fmt"
"io"
"net/http"
"strings"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common"
"github.com/songquanpeng/one-api/common/client"
@@ -21,9 +25,6 @@ import (
"github.com/songquanpeng/one-api/relay/meta"
relaymodel "github.com/songquanpeng/one-api/relay/model"
"github.com/songquanpeng/one-api/relay/relaymode"
"io"
"net/http"
"strings"
)
func RelayAudioHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatusCode {
@@ -53,7 +54,7 @@ func RelayAudioHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
}
}
modelRatio := billingratio.GetModelRatio(audioModel)
modelRatio := billingratio.GetModelRatio(audioModel, channelType)
groupRatio := billingratio.GetGroupRatio(group)
ratio := modelRatio * groupRatio
var quota int64

9
relay/controller/helper.go
View File

@@ -4,6 +4,10 @@ import (
"context"
"errors"
"fmt"
"math"
"net/http"
"strings"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common"
"github.com/songquanpeng/one-api/common/config"
@@ -16,9 +20,6 @@ import (
"github.com/songquanpeng/one-api/relay/meta"
relaymodel "github.com/songquanpeng/one-api/relay/model"
"github.com/songquanpeng/one-api/relay/relaymode"
"math"
"net/http"
"strings"
)
func getAndValidateTextRequest(c *gin.Context, relayMode int) (*relaymodel.GeneralOpenAIRequest, error) {
@@ -95,7 +96,7 @@ func postConsumeQuota(ctx context.Context, usage *relaymodel.Usage, meta *meta.M
return
}
var quota int64
completionRatio := billingratio.GetCompletionRatio(textRequest.Model)
completionRatio := billingratio.GetCompletionRatio(textRequest.Model, meta.ChannelType)
promptTokens := usage.PromptTokens
completionTokens := usage.CompletionTokens
quota = int64(math.Ceil((float64(promptTokens) + float64(completionTokens)*completionRatio) * ratio))

7
relay/controller/image.go
View File

@@ -6,6 +6,9 @@ import (
"encoding/json"
"errors"
"fmt"
"io"
"net/http"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common"
"github.com/songquanpeng/one-api/common/ctxkey"
@@ -17,8 +20,6 @@ import (
"github.com/songquanpeng/one-api/relay/channeltype"
"github.com/songquanpeng/one-api/relay/meta"
relaymodel "github.com/songquanpeng/one-api/relay/model"
"io"
"net/http"
)
func getImageRequest(c *gin.Context, relayMode int) (*relaymodel.ImageRequest, error) {
@@ -166,7 +167,7 @@ func RelayImageHelper(c *gin.Context, relayMode int) *relaymodel.ErrorWithStatus
requestBody = bytes.NewBuffer(jsonStr)
}
modelRatio := billingratio.GetModelRatio(imageModel)
modelRatio := billingratio.GetModelRatio(imageModel, meta.ChannelType)
groupRatio := billingratio.GetGroupRatio(meta.Group)
ratio := modelRatio * groupRatio
userQuota, err := model.CacheGetUserQuota(ctx, meta.UserId)

7
relay/controller/text.go
View File

@@ -4,6 +4,9 @@ import (
"bytes"
"encoding/json"
"fmt"
"io"
"net/http"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common/logger"
"github.com/songquanpeng/one-api/relay"
@@ -14,8 +17,6 @@ import (
"github.com/songquanpeng/one-api/relay/channeltype"
"github.com/songquanpeng/one-api/relay/meta"
"github.com/songquanpeng/one-api/relay/model"
"io"
"net/http"
)
func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
@@ -35,7 +36,7 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
textRequest.Model, isModelMapped = getMappedModelName(textRequest.Model, meta.ModelMapping)
meta.ActualModelName = textRequest.Model
// get model ratio & group ratio
modelRatio := billingratio.GetModelRatio(textRequest.Model)
modelRatio := billingratio.GetModelRatio(textRequest.Model, meta.ChannelType)
groupRatio := billingratio.GetGroupRatio(meta.Group)
ratio := modelRatio * groupRatio
// pre-consume quota