refactor: remove reka client (#635)

pull/636/head
sigoden 2 weeks ago committed by GitHub
parent 52a847743e
commit 590c525048
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

@ -229,27 +229,6 @@ models-cohere() {
}
# @cmd Chat with reka api
# @env REKA_API_KEY!
# @option -m --model=reka-flash $REKA_MODEL
# @flag -S --no-stream
# @arg text~
chat-reka() {
_wrapper curl -i https://api.reka.ai/v1/chat \
-X POST \
-H 'Content-Type: application/json' \
-H "X-Api-Key: $REKA_API_KEY" \
-d "$(_build_body reka "$@")"
}
# @cmd List reka models
# @env REKA_API_KEY!
models-reka() {
_wrapper curl https://api.reka.ai/v1/models \
-H "X-Api-Key: $REKA_API_KEY" \
}
# @cmd Chat with ollama api
# @option -m --model=codegemma $OLLAMA_MODEL
# @flag -S --no-stream
@ -492,18 +471,6 @@ _build_body() {
"model": "'$argc_model'",
"message": "'"$*"'",
"stream": '$stream'
}'
;;
reka)
echo '{
"model": "'$argc_model'",
"messages": [
{
"role": "user",
"content": "'"$*"'"
}
],
"stream": '$stream'
}'
;;
claude)

@ -31,7 +31,6 @@ AIChat is an all-in-one AI CLI tool featuring Chat-REPL, RAG, Function Calling,
- Claude: Claude-3.5/Claude-3 (paid, vision, function-calling)
- Mistral (paid, embedding, function-calling)
- Cohere: Command-R/Command-R+ (paid, embedding, rerank, function-calling)
- Reka (paid, vision)
- Perplexity: Llama-3/Mixtral (paid)
- Groq: Llama-3/Mixtral/Gemma (free, function-calling)
- Ollama (free, local, embedding)

@ -144,10 +144,6 @@ clients:
- type: cohere
api_key: xxx # ENV: {client}_API_KEY
# See https://docs.reka.ai/quick-start
- type: reka
api_key: xxx # ENV: {client}_API_KEY
# See https://docs.perplexity.ai/docs/getting-started
- type: openai-compatible
name: perplexity

@ -189,28 +189,6 @@
type: rerank
max_input_tokens: 4096
- platform: reka
docs:
# - https://www.reka.ai/ourmodels
# - https://www.reka.ai/reka-deploy
# - https://docs.reka.ai/api-reference/chat/create
models:
- name: reka-core
max_input_tokens: 128000
input_price: 10
output_price: 25
supports_vision: true
- name: reka-flash
max_input_tokens: 128000
input_price: 0.8
output_price: 2
supports_vision: true
- name: reka-edge
max_input_tokens: 128000
input_price: 0.4
output_price: 1
supports_vision: true
- platform: perplexity
# docs:
# - https://docs.perplexity.ai/docs/model-cards

@ -24,7 +24,6 @@ register_client!(
(gemini, "gemini", GeminiConfig, GeminiClient),
(claude, "claude", ClaudeConfig, ClaudeClient),
(cohere, "cohere", CohereConfig, CohereClient),
(reka, "reka", RekaConfig, RekaClient),
(ollama, "ollama", OllamaConfig, OllamaClient),
(
azure_openai,

@ -1,124 +0,0 @@
use super::*;
use anyhow::{anyhow, Result};
use reqwest::{Client as ReqwestClient, RequestBuilder};
use serde::Deserialize;
use serde_json::{json, Value};
const CHAT_COMPLETIONS_API_URL: &str = "https://api.reka.ai/v1/chat";
#[derive(Debug, Clone, Deserialize, Default)]
pub struct RekaConfig {
pub name: Option<String>,
pub api_key: Option<String>,
#[serde(default)]
pub models: Vec<ModelData>,
pub patches: Option<ModelPatches>,
pub extra: Option<ExtraConfig>,
}
impl RekaClient {
config_get_fn!(api_key, get_api_key);
pub const PROMPTS: [PromptAction<'static>; 1] =
[("api_key", "API Key:", true, PromptKind::String)];
fn chat_completions_builder(
&self,
client: &ReqwestClient,
data: ChatCompletionsData,
) -> Result<RequestBuilder> {
let api_key = self.get_api_key()?;
let mut body = build_chat_completions_body(data, &self.model);
self.patch_chat_completions_body(&mut body);
let url = CHAT_COMPLETIONS_API_URL;
debug!("Reka Chat Completions Request: {url} {body}");
let builder = client.post(url).header("x-api-key", api_key).json(&body);
Ok(builder)
}
}
impl_client_trait!(RekaClient, chat_completions, chat_completions_streaming);
async fn chat_completions(builder: RequestBuilder) -> Result<ChatCompletionsOutput> {
let res = builder.send().await?;
let status = res.status();
let data: Value = res.json().await?;
if !status.is_success() {
catch_error(&data, status.as_u16())?;
}
debug!("non-stream-data: {data}");
extract_chat_completions(&data)
}
async fn chat_completions_streaming(
builder: RequestBuilder,
handler: &mut SseHandler,
) -> Result<()> {
let mut prev_text = String::new();
let handle = |message: SseMmessage| -> Result<bool> {
let data: Value = serde_json::from_str(&message.data)?;
debug!("stream-data: {data}");
if let Some(text) = data["responses"][0]["chunk"]["content"].as_str() {
let delta_text = &text[prev_text.len()..];
prev_text = text.to_string();
handler.text(delta_text)?;
}
Ok(false)
};
sse_stream(builder, handle).await
}
fn build_chat_completions_body(data: ChatCompletionsData, model: &Model) -> Value {
let ChatCompletionsData {
mut messages,
temperature,
top_p,
functions: _,
stream,
} = data;
patch_system_message(&mut messages);
let mut body = json!({
"model": &model.name(),
"messages": messages,
});
if let Some(v) = model.max_tokens_param() {
body["max_tokens"] = v.into();
}
if let Some(v) = temperature {
body["temperature"] = v.into();
}
if let Some(v) = top_p {
body["top_p"] = v.into();
}
if stream {
body["stream"] = true.into();
}
body
}
fn extract_chat_completions(data: &Value) -> Result<ChatCompletionsOutput> {
let text = data["responses"][0]["message"]["content"]
.as_str()
.ok_or_else(|| anyhow!("Invalid response data: {data}"))?;
let output = ChatCompletionsOutput {
text: text.to_string(),
tool_calls: vec![],
id: data["id"].as_str().map(|v| v.to_string()),
input_tokens: data["usage"]["input_tokens"].as_u64(),
output_tokens: data["usage"]["output_tokens"].as_u64(),
};
Ok(output)
}
Loading…
Cancel
Save