gpt4free/g4f/models.py

226 lines
5.0 KiB
Python
Raw Normal View History

2023-07-25 00:56:49 +00:00
from dataclasses import dataclass
2023-06-24 01:47:00 +00:00
from .Provider import Bard, BaseProvider, GetGpt, H2o, Liaobots, Vercel, Equing
2023-07-28 10:07:17 +00:00
2023-06-24 01:47:00 +00:00
2023-07-25 00:56:49 +00:00
@dataclass
2023-06-24 01:47:00 +00:00
class Model:
2023-07-25 00:56:49 +00:00
name: str
base_provider: str
2023-07-28 10:07:17 +00:00
best_provider: type[BaseProvider]
2023-07-25 00:56:49 +00:00
2023-07-28 10:07:17 +00:00
# GPT-3.5 / GPT-4
2023-07-25 00:56:49 +00:00
gpt_35_turbo = Model(
name="gpt-3.5-turbo",
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=GetGpt,
2023-07-25 00:56:49 +00:00
)
gpt_4 = Model(
name="gpt-4",
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Liaobots,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
# Bard
palm = Model(
name="palm",
base_provider="google",
best_provider=Bard,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
# H2o
falcon_7b = Model(
name="h2oai/h2ogpt-gm-oasst1-en-2048-falcon-7b-v3",
base_provider="huggingface",
best_provider=H2o,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
falcon_40b = Model(
name="h2oai/h2ogpt-gm-oasst1-en-2048-falcon-40b-v1",
base_provider="huggingface",
best_provider=H2o,
)
llama_13b = Model(
name="h2oai/h2ogpt-gm-oasst1-en-2048-open-llama-13b",
base_provider="huggingface",
best_provider=H2o,
)
# Vercel
claude_instant_v1 = Model(
name="anthropic:claude-instant-v1",
2023-07-25 00:56:49 +00:00
base_provider="anthropic",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
claude_v1 = Model(
2023-07-28 10:07:17 +00:00
name="anthropic:claude-v1",
2023-07-25 00:56:49 +00:00
base_provider="anthropic",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
claude_v2 = Model(
name="anthropic:claude-v2",
base_provider="anthropic",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
command_light_nightly = Model(
name="cohere:command-light-nightly",
base_provider="cohere",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
command_nightly = Model(
name="cohere:command-nightly",
base_provider="cohere",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
gpt_neox_20b = Model(
name="huggingface:EleutherAI/gpt-neox-20b",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
oasst_sft_1_pythia_12b = Model(
name="huggingface:OpenAssistant/oasst-sft-1-pythia-12b",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
oasst_sft_4_pythia_12b_epoch_35 = Model(
name="huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
santacoder = Model(
name="huggingface:bigcode/santacoder",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
bloom = Model(
name="huggingface:bigscience/bloom",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
flan_t5_xxl = Model(
name="huggingface:google/flan-t5-xxl",
2023-07-25 00:56:49 +00:00
base_provider="huggingface",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
code_davinci_002 = Model(
name="openai:code-davinci-002",
base_provider="openai",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
gpt_35_turbo_16k = Model(
name="openai:gpt-3.5-turbo-16k",
base_provider="openai",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
gpt_35_turbo_16k_0613 = Model(
name="openai:gpt-3.5-turbo-16k-0613",
2023-07-25 00:56:49 +00:00
base_provider="openai",
best_provider=Equing,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
gpt_4_0613 = Model(
name="openai:gpt-4-0613",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
text_ada_001 = Model(
2023-07-28 10:07:17 +00:00
name="openai:text-ada-001",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
text_babbage_001 = Model(
2023-07-28 10:07:17 +00:00
name="openai:text-babbage-001",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
text_curie_001 = Model(
2023-07-28 10:07:17 +00:00
name="openai:text-curie-001",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
text_davinci_002 = Model(
2023-07-28 10:07:17 +00:00
name="openai:text-davinci-002",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
text_davinci_003 = Model(
2023-07-28 10:07:17 +00:00
name="openai:text-davinci-003",
2023-07-25 00:56:49 +00:00
base_provider="openai",
2023-07-28 10:07:17 +00:00
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
llama13b_v2_chat = Model(
name="replicate:a16z-infra/llama13b-v2-chat",
base_provider="replicate",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-07-28 10:07:17 +00:00
llama7b_v2_chat = Model(
name="replicate:a16z-infra/llama7b-v2-chat",
base_provider="replicate",
best_provider=Vercel,
2023-07-25 00:56:49 +00:00
)
2023-06-24 01:47:00 +00:00
class ModelUtils:
2023-07-25 00:56:49 +00:00
convert: dict[str, Model] = {
2023-07-28 10:07:17 +00:00
# GPT-3.5 / GPT-4
2023-07-25 00:56:49 +00:00
"gpt-3.5-turbo": gpt_35_turbo,
"gpt-4": gpt_4,
2023-07-28 10:07:17 +00:00
# Bard
"palm2": palm,
"palm": palm,
"google": palm,
"google-bard": palm,
"google-palm": palm,
"bard": palm,
# H2o
"falcon-40b": falcon_40b,
"falcon-7b": falcon_7b,
"llama-13b": llama_13b,
# Vercel
2023-07-25 00:56:49 +00:00
"claude-instant-v1": claude_instant_v1,
"claude-v1": claude_v1,
2023-07-28 10:07:17 +00:00
"claude-v2": claude_v2,
"command-light-nightly": command_light_nightly,
"command-nightly": command_nightly,
2023-07-25 00:56:49 +00:00
"gpt-neox-20b": gpt_neox_20b,
2023-07-28 10:07:17 +00:00
"oasst-sft-1-pythia-12b": oasst_sft_1_pythia_12b,
2023-07-25 00:56:49 +00:00
"oasst-sft-4-pythia-12b-epoch-3.5": oasst_sft_4_pythia_12b_epoch_35,
"santacoder": santacoder,
2023-07-28 10:07:17 +00:00
"bloom": bloom,
"flan-t5-xxl": flan_t5_xxl,
2023-07-25 00:56:49 +00:00
"code-davinci-002": code_davinci_002,
2023-07-28 10:07:17 +00:00
"gpt-3.5-turbo-16k": gpt_35_turbo_16k,
"gpt-3.5-turbo-16k-0613": gpt_35_turbo_16k_0613,
"gpt-4-0613": gpt_4_0613,
2023-07-25 00:56:49 +00:00
"text-ada-001": text_ada_001,
"text-babbage-001": text_babbage_001,
"text-curie-001": text_curie_001,
"text-davinci-002": text_davinci_002,
"text-davinci-003": text_davinci_003,
2023-07-28 10:07:17 +00:00
"llama13b-v2-chat": llama13b_v2_chat,
"llama7b-v2-chat": llama7b_v2_chat,
2023-07-25 00:56:49 +00:00
}