From efd75a11b871d61ac31b0e274acdfb33daba361d Mon Sep 17 00:00:00 2001 From: abc <98614666+xtekky@users.noreply.github.com> Date: Sun, 27 Aug 2023 17:37:44 +0200 Subject: ~ | code styling --- g4f/models.py | 270 ++++++++++++++++++++++++++-------------------------------- 1 file changed, 122 insertions(+), 148 deletions(-) (limited to 'g4f/models.py') diff --git a/g4f/models.py b/g4f/models.py index 64951792..e61ebbcd 100644 --- a/g4f/models.py +++ b/g4f/models.py @@ -1,8 +1,6 @@ from dataclasses import dataclass - from .Provider import Bard, BaseProvider, GetGpt, H2o, Liaobots, Vercel, Equing - @dataclass class Model: name: str @@ -12,214 +10,190 @@ class Model: # GPT-3.5 / GPT-4 gpt_35_turbo = Model( - name="gpt-3.5-turbo", - base_provider="openai", - best_provider=GetGpt, -) + name = 'gpt-3.5-turbo', + base_provider = 'openai', + best_provider = GetGpt) gpt_4 = Model( - name="gpt-4", - base_provider="openai", - best_provider=Liaobots, -) + name = 'gpt-4', + base_provider = 'openai', + best_provider = Liaobots) # Bard palm = Model( - name="palm", - base_provider="google", - best_provider=Bard, -) + name = 'palm', + base_provider = 'google', + best_provider = Bard) # H2o falcon_7b = Model( - name="h2oai/h2ogpt-gm-oasst1-en-2048-falcon-7b-v3", - base_provider="huggingface", - best_provider=H2o, -) + name = 'h2oai/h2ogpt-gm-oasst1-en-2048-falcon-7b-v3', + base_provider = 'huggingface', + best_provider = H2o) falcon_40b = Model( - name="h2oai/h2ogpt-gm-oasst1-en-2048-falcon-40b-v1", - base_provider="huggingface", - best_provider=H2o, -) + 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, -) + 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=Vercel, -) + name = 'anthropic:claude-instant-v1', + base_provider = 'anthropic', + best_provider = Vercel) claude_v1 = Model( - name="anthropic:claude-v1", - base_provider="anthropic", - best_provider=Vercel, -) + name = 'anthropic:claude-v1', + base_provider = 'anthropic', + best_provider = Vercel) claude_v2 = Model( - name="anthropic:claude-v2", - base_provider="anthropic", - best_provider=Vercel, -) + name = 'anthropic:claude-v2', + base_provider = 'anthropic', + best_provider = Vercel) command_light_nightly = Model( - name="cohere:command-light-nightly", - base_provider="cohere", - best_provider=Vercel, -) + name = 'cohere:command-light-nightly', + base_provider = 'cohere', + best_provider = Vercel) command_nightly = Model( - name="cohere:command-nightly", - base_provider="cohere", - best_provider=Vercel, -) + name = 'cohere:command-nightly', + base_provider = 'cohere', + best_provider = Vercel) gpt_neox_20b = Model( - name="huggingface:EleutherAI/gpt-neox-20b", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:EleutherAI/gpt-neox-20b', + base_provider = 'huggingface', + best_provider = Vercel) oasst_sft_1_pythia_12b = Model( - name="huggingface:OpenAssistant/oasst-sft-1-pythia-12b", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:OpenAssistant/oasst-sft-1-pythia-12b', + base_provider = 'huggingface', + best_provider = Vercel) oasst_sft_4_pythia_12b_epoch_35 = Model( - name="huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5', + base_provider = 'huggingface', + best_provider = Vercel) santacoder = Model( - name="huggingface:bigcode/santacoder", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:bigcode/santacoder', + base_provider = 'huggingface', + best_provider = Vercel) bloom = Model( - name="huggingface:bigscience/bloom", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:bigscience/bloom', + base_provider = 'huggingface', + best_provider = Vercel) flan_t5_xxl = Model( - name="huggingface:google/flan-t5-xxl", - base_provider="huggingface", - best_provider=Vercel, -) + name = 'huggingface:google/flan-t5-xxl', + base_provider = 'huggingface', + best_provider = Vercel) code_davinci_002 = Model( - name="openai:code-davinci-002", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:code-davinci-002', + base_provider = 'openai', + best_provider = Vercel) gpt_35_turbo_16k = Model( - name="openai:gpt-3.5-turbo-16k", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:gpt-3.5-turbo-16k', + base_provider = 'openai', + best_provider = Vercel) gpt_35_turbo_16k_0613 = Model( - name="openai:gpt-3.5-turbo-16k-0613", - base_provider="openai", - best_provider=Equing, -) + name = 'openai:gpt-3.5-turbo-16k-0613', + base_provider = 'openai', + best_provider = Equing) gpt_4_0613 = Model( - name="openai:gpt-4-0613", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:gpt-4-0613', + base_provider = 'openai', + best_provider = Vercel) text_ada_001 = Model( - name="openai:text-ada-001", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:text-ada-001', + base_provider = 'openai', + best_provider = Vercel) text_babbage_001 = Model( - name="openai:text-babbage-001", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:text-babbage-001', + base_provider = 'openai', + best_provider = Vercel) text_curie_001 = Model( - name="openai:text-curie-001", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:text-curie-001', + base_provider = 'openai', + best_provider = Vercel) text_davinci_002 = Model( - name="openai:text-davinci-002", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:text-davinci-002', + base_provider = 'openai', + best_provider = Vercel) text_davinci_003 = Model( - name="openai:text-davinci-003", - base_provider="openai", - best_provider=Vercel, -) + name = 'openai:text-davinci-003', + base_provider = 'openai', + best_provider = Vercel) llama13b_v2_chat = Model( - name="replicate:a16z-infra/llama13b-v2-chat", - base_provider="replicate", - best_provider=Vercel, -) + name = 'replicate:a16z-infra/llama13b-v2-chat', + base_provider = 'replicate', + best_provider = Vercel) llama7b_v2_chat = Model( - name="replicate:a16z-infra/llama7b-v2-chat", - base_provider="replicate", - best_provider=Vercel, -) + name = 'replicate:a16z-infra/llama7b-v2-chat', + base_provider = 'replicate', + best_provider = Vercel) class ModelUtils: convert: dict[str, Model] = { # GPT-3.5 / GPT-4 - "gpt-3.5-turbo": gpt_35_turbo, - "gpt-4": gpt_4, + 'gpt-3.5-turbo' : gpt_35_turbo, + 'gpt-4' : gpt_4, + # Bard - "palm2": palm, - "palm": palm, - "google": palm, - "google-bard": palm, - "google-palm": palm, - "bard": palm, + '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, + 'falcon-40b' : falcon_40b, + 'falcon-7b' : falcon_7b, + 'llama-13b' : llama_13b, + # Vercel - "claude-instant-v1": claude_instant_v1, - "claude-v1": claude_v1, - "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, - "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, - "llama13b-v2-chat": llama13b_v2_chat, - "llama7b-v2-chat": llama7b_v2_chat, - } + 'claude-instant-v1' : claude_instant_v1, + 'claude-v1' : claude_v1, + 'claude-v2' : claude_v2, + 'command-nightly' : command_nightly, + 'gpt-neox-20b' : gpt_neox_20b, + 'santacoder' : santacoder, + 'bloom' : bloom, + 'flan-t5-xxl' : flan_t5_xxl, + 'code-davinci-002' : code_davinci_002, + 'gpt-3.5-turbo-16k' : gpt_35_turbo_16k, + '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, + 'llama13b-v2-chat' : llama13b_v2_chat, + 'llama7b-v2-chat' : llama7b_v2_chat, + + '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, + 'command-light-nightly' : command_light_nightly, + 'gpt-3.5-turbo-16k-0613' : gpt_35_turbo_16k_0613, + } \ No newline at end of file -- cgit v1.2.3