kaz-llm-lb / src /gen /config /api_config.yaml
hi-melnikov's picture
First setup of leaderboard
8e67ebe
raw
history blame
5.33 kB
# name: str
# model_name: str
# endpoints: default to null
# - api_base: str
# api_key: str optional (required if no api_key_ENV)
# api_key_ENV: str optional (ENV name to store the token secret)
# api_version: str optional (only for azure)
# api_type: str
# tokenizer: str optional (to optimize token limits)
# parallel: int
gpt-4-1106-preview:
model_name: gpt-4-1106-preview
endpoints:
- api_base: https://cgiaura-openai-trainning.openai.azure.com
api_key_ENV: GPT_4_TOKEN
api_version: 2024-02-15-preview
api_type: azure
parallel: 5
gpt-3.5-turbo-0125:
model_name: gpt-3.5-turbo-0125
endpoints:
- api_base: https://api.openai.com/v1/
api_key_ENV: GPT_3_TOKEN
api_type: openai
parallel: 6
gpt-3.5-turbo-0125-ru-sys:
model_name: gpt-3.5-turbo-0125
endpoints:
- api_base: https://api.openai.com/v1/
api_key_ENV: GPT_3_TOKEN
system_prompt: You are a helpful assistant. Answer on Russian.
api_type: openai
parallel: 6
yandex_gpt_pro:
model_name: yandexgpt
endpoints:
- catalog_id: b1gk1i41eeb97a5s68c7
iam_token_ENV: YANDEX_GPT_TOKEN
api_type: yandex
parallel: 2
gigachat_lite:
model_name: GigaChat
endpoints:
auth_token_ENV: GIGACHAT_GPT_TOKEN
api_type: gigachat
parallel: 1
gigachat_pro:
model_name: GigaChat-Pro
endpoints:
auth_token_ENV: GIGACHAT_GPT_TOKEN
api_type: gigachat
parallel: 1
meta-llama-3-70b-instruct-gptq:
model_name: MaziyarPanahi/Meta-Llama-3-70B-Instruct-GPTQ
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
snorkel-mistral-pairrm-dpo:
model_name: snorkelai/Snorkel-Mistral-PairRM-DPO
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
sfr-iterative-dpo-llama-3-8b-r:
model_name: Salesforce/SFR-Iterative-DPO-LLaMA-3-8B-R
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
openchat-3.5-0106:
model_name: openchat/openchat-3.5-0106
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
mixtral-8x7b-instruct-v0.1:
model_name: LoneStriker/Mixtral-8x7B-Instruct-v0.1-HF
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 4
neural-chat-7b-v3-3:
model_name: Intel/neural-chat-7b-v3-3
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
meta-llama-3-8b-instruct:
model_name: meta-llama/Meta-Llama-3-8B-Instruct
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
saiga_llama3_8b:
model_name: IlyaGusev/saiga_llama3_8b
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
hermes-2-pro-llama-3-8b:
model_name: NousResearch/Hermes-2-Pro-Llama-3-8B
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
dpopenhermes-7b:
model_name: openaccess-ai-collective/DPOpenHermes-7B
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
llama3-chatqa-1.5-8b:
model_name: nvidia/Llama3-ChatQA-1.5-8B
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
hermes-2-pro-mistral-7b:
model_name: NousResearch/Hermes-2-Pro-Mistral-7B
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
suzume-llama-3-8b-multilingual:
model_name: lightblue/suzume-llama-3-8B-multilingual
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
vikhr-7b-instruct_0.4:
model_name: Vikhrmodels/Vikhr-7B-instruct_0.4
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
vikhr-it-5.2-fp16-cp:
model_name: Vikhrmodels/it-5.2-fp16-cp
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
system_prompt: Ты Вихрь, русскоязычный ассистент.
parallel: 6
starling-lm-7b-beta:
model_name: Nexusflow/Starling-LM-7B-beta
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
c4ai-command-r-v01:
model_name: CohereForAI/c4ai-command-r-v01
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 6
starcoder2-15b-instruct-v0.1:
model_name: bigcode/starcoder2-15b-instruct-v0.1
endpoints:
- api_base: http://localhost:8000/v1
api_key: token-abc123
api_type: openai
parallel: 3