summaryrefslogtreecommitdiffstats
path: root/g4f/models.py
diff options
context:
space:
mode:
Diffstat (limited to 'g4f/models.py')
-rw-r--r--g4f/models.py260
1 files changed, 121 insertions, 139 deletions
diff --git a/g4f/models.py b/g4f/models.py
index 27336244..40d21f2a 100644
--- a/g4f/models.py
+++ b/g4f/models.py
@@ -1,243 +1,225 @@
-from types import ModuleType
-from . import Provider
from dataclasses import dataclass
+from .provider import Bard, BaseProvider, GetGpt, H2o, Liaobots, Vercel
+
@dataclass
class Model:
name: str
base_provider: str
- best_provider: ModuleType or None
+ best_provider: type[BaseProvider]
+
+# GPT-3.5 / GPT-4
gpt_35_turbo = Model(
name="gpt-3.5-turbo",
base_provider="openai",
- best_provider=Provider.GetGpt,
+ best_provider=GetGpt,
)
gpt_4 = Model(
name="gpt-4",
base_provider="openai",
- best_provider=Provider.Bing,
+ best_provider=Liaobots,
)
-claude_instant_v1_100k = Model(
- name="claude-instant-v1-100k",
- base_provider="anthropic",
- best_provider=Provider.Vercel,
+# Bard
+palm = Model(
+ name="palm",
+ base_provider="google",
+ best_provider=Bard,
)
-claude_instant_v1 = Model(
- name="claude-instant-v1",
- base_provider="anthropic",
- best_provider=Provider.Vercel,
+# H2o
+falcon_7b = Model(
+ name="h2oai/h2ogpt-gm-oasst1-en-2048-falcon-7b-v3",
+ base_provider="huggingface",
+ best_provider=H2o,
)
-claude_v1_100k = Model(
- name="claude-v1-100k",
+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",
base_provider="anthropic",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
claude_v1 = Model(
- name="claude-v1",
+ name="anthropic:claude-v1",
base_provider="anthropic",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-alpaca_7b = Model(
- name="alpaca-7b",
- base_provider="replicate",
- best_provider=Provider.Vercel,
+claude_v2 = Model(
+ name="anthropic:claude-v2",
+ base_provider="anthropic",
+ best_provider=Vercel,
)
-stablelm_tuned_alpha_7b = Model(
- name="stablelm-tuned-alpha-7b",
- base_provider="replicate",
- best_provider=Provider.Vercel,
+command_light_nightly = Model(
+ name="cohere:command-light-nightly",
+ base_provider="cohere",
+ best_provider=Vercel,
)
-bloom = Model(
- name="bloom",
- base_provider="huggingface",
- best_provider=Provider.Vercel,
+command_nightly = Model(
+ name="cohere:command-nightly",
+ base_provider="cohere",
+ best_provider=Vercel,
)
-bloomz = Model(
- name="bloomz",
+gpt_neox_20b = Model(
+ name="huggingface:EleutherAI/gpt-neox-20b",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-flan_t5_xxl = Model(
- name="flan-t5-xxl",
+oasst_sft_1_pythia_12b = Model(
+ name="huggingface:OpenAssistant/oasst-sft-1-pythia-12b",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-flan_ul2 = Model(
- name="flan-ul2",
+oasst_sft_4_pythia_12b_epoch_35 = Model(
+ name="huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-gpt_neox_20b = Model(
- name="gpt-neox-20b",
+santacoder = Model(
+ name="huggingface:bigcode/santacoder",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-oasst_sft_4_pythia_12b_epoch_35 = Model(
- name="oasst-sft-4-pythia-12b-epoch-3.5",
+bloom = Model(
+ name="huggingface:bigscience/bloom",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-santacoder = Model(
- name="santacoder",
+flan_t5_xxl = Model(
+ name="huggingface:google/flan-t5-xxl",
base_provider="huggingface",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-command_medium_nightly = Model(
- name="command-medium-nightly",
- base_provider="cohere",
- best_provider=Provider.Vercel,
+code_davinci_002 = Model(
+ name="openai:code-davinci-002",
+ base_provider="openai",
+ best_provider=Vercel,
)
-command_xlarge_nightly = Model(
- name="command-xlarge-nightly",
- base_provider="cohere",
- best_provider=Provider.Vercel,
+gpt_35_turbo_16k = Model(
+ name="openai:gpt-3.5-turbo-16k",
+ base_provider="openai",
+ best_provider=Vercel,
)
-code_cushman_001 = Model(
- name="code-cushman-001",
+gpt_35_turbo_16k_0613 = Model(
+ name="openai:gpt-3.5-turbo-16k-0613",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
-code_davinci_002 = Model(
- name="code-davinci-002",
+gpt_4_0613 = Model(
+ name="openai:gpt-4-0613",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
text_ada_001 = Model(
- name="text-ada-001",
+ name="openai:text-ada-001",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
text_babbage_001 = Model(
- name="text-babbage-001",
+ name="openai:text-babbage-001",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
text_curie_001 = Model(
- name="text-curie-001",
+ name="openai:text-curie-001",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
text_davinci_002 = Model(
- name="text-davinci-002",
+ name="openai:text-davinci-002",
base_provider="openai",
- best_provider=Provider.Vercel,
+ best_provider=Vercel,
)
text_davinci_003 = Model(
- name="text-davinci-003",
+ name="openai:text-davinci-003",
base_provider="openai",
- best_provider=Provider.Vercel,
-)
-
-palm = Model(
- name="palm",
- base_provider="google",
- best_provider=Provider.Bard,
+ best_provider=Vercel,
)
-falcon_40b = Model(
- name="falcon-40b",
- base_provider="huggingface",
- best_provider=Provider.H2o,
-)
-
-falcon_7b = Model(
- name="falcon-7b",
- base_provider="huggingface",
- best_provider=Provider.H2o,
-)
-
-llama_13b = Model(
- name="llama-13b",
- base_provider="huggingface",
- best_provider=Provider.H2o,
-)
-
-gpt_35_turbo_16k = Model(
- name="gpt-3.5-turbo-16k",
- base_provider="openai",
- best_provider=Provider.EasyChat,
-)
-
-gpt_35_turbo_0613 = Model(
- name="gpt-3.5-turbo-0613",
- base_provider="openai",
- best_provider=Provider.EasyChat,
+llama13b_v2_chat = Model(
+ name="replicate:a16z-infra/llama13b-v2-chat",
+ base_provider="replicate",
+ best_provider=Vercel,
)
-gpt_35_turbo_16k_0613 = Model(
- name="gpt-3.5-turbo-16k-0613",
- base_provider="openai",
- best_provider=Provider.EasyChat,
+llama7b_v2_chat = Model(
+ name="replicate:a16z-infra/llama7b-v2-chat",
+ base_provider="replicate",
+ best_provider=Vercel,
)
-gpt_4_32k = Model(name="gpt-4-32k", base_provider="openai", best_provider=None)
-
-gpt_4_0613 = Model(name="gpt-4-0613", base_provider="openai", best_provider=None)
-
class ModelUtils:
convert: dict[str, Model] = {
+ # GPT-3.5 / GPT-4
"gpt-3.5-turbo": gpt_35_turbo,
- "gpt-3.5-turbo-16k": gpt_35_turbo_16k,
- "gpt-3.5-turbo-0613": gpt_35_turbo_0613,
- "gpt-3.5-turbo-16k-0613": gpt_35_turbo_16k_0613,
"gpt-4": gpt_4,
- "gpt-4-32k": gpt_4_32k,
- "gpt-4-0613": gpt_4_0613,
- "claude-instant-v1-100k": claude_instant_v1_100k,
- "claude-v1-100k": claude_v1_100k,
+ # 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
"claude-instant-v1": claude_instant_v1,
"claude-v1": claude_v1,
- "alpaca-7b": alpaca_7b,
- "stablelm-tuned-alpha-7b": stablelm_tuned_alpha_7b,
- "bloom": bloom,
- "bloomz": bloomz,
- "flan-t5-xxl": flan_t5_xxl,
- "flan-ul2": flan_ul2,
+ "claude-v2": claude_v2,
+ "command-light-nightly": command_light_nightly,
+ "command-nightly": command_nightly,
"gpt-neox-20b": gpt_neox_20b,
+ "oasst-sft-1-pythia-12b": oasst_sft_1_pythia_12b,
"oasst-sft-4-pythia-12b-epoch-3.5": oasst_sft_4_pythia_12b_epoch_35,
"santacoder": santacoder,
- "command-medium-nightly": command_medium_nightly,
- "command-xlarge-nightly": command_xlarge_nightly,
- "code-cushman-001": code_cushman_001,
+ "bloom": bloom,
+ "flan-t5-xxl": flan_t5_xxl,
"code-davinci-002": code_davinci_002,
+ "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,
"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,
- "palm2": palm,
- "palm": palm,
- "google": palm,
- "google-bard": palm,
- "google-palm": palm,
- "bard": palm,
- "falcon-40b": falcon_40b,
- "falcon-7b": falcon_7b,
- "llama-13b": llama_13b,
+ "llama13b-v2-chat": llama13b_v2_chat,
+ "llama7b-v2-chat": llama7b_v2_chat,
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-26 09:37:53 +0200