Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

💄 style: add deepseek r1 distill models for qwen series #5850

Merged
merged 7 commits into from
Feb 15, 2025
Merged

💄 style: add deepseek r1 distill models for qwen series #5850

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
6f1ac3c
feat: add deepseek distill models for qwen series
liyuan-rey Feb 7, 2025
b25c393
Merge branch 'main' into feat/deepseek-r1-distill-qwen
liyuan-rey Feb 8, 2025
d1802be
style: update qwen.ts indentation
liyuan-rey Feb 8, 2025
c48f860
Merge branch 'lobehub:main' into feat/deepseek-r1-distill-qwen
liyuan-rey Feb 8, 2025
c79fff9
style: update qwen aiModels and modelProviders
liyuan-rey Feb 8, 2025
8f6530e
Merge branch 'main' into feat/deepseek-r1-distill-qwen
liyuan-rey Feb 10, 2025
c33c4c7
Merge branch 'main' into feat/deepseek-r1-distill-qwen
liyuan-rey Feb 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 18 additions & 18 deletions src/config/aiModels/qwen.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -376,7 +376,7 @@ const qwenChatModels: AIChatModelCard[] = [
vision: true,
},
contextWindowTokens: 131_072,
description:
description:
'指令跟随、数学、解题、代码整体提升，万物识别能力提升，支持多样格式直接精准定位视觉元素，支持对长视频文件（最长10分钟）进行理解和秒级别的事件时刻定位，能理解时间先后和快慢，基于解析和定位能力支持操控OS或Mobile的Agent，关键信息抽取能力和Json格式输出能力强，此版本为72B版本，本系列能力最强的版本。',
displayName: 'Qwen2.5 VL 72B',
id: 'qwen2.5-vl-72b-instruct',
Expand All @@ -394,7 +394,7 @@ const qwenChatModels: AIChatModelCard[] = [
vision: true,
},
contextWindowTokens: 131_072,
description:
description:
'指令跟随、数学、解题、代码整体提升，万物识别能力提升，支持多样格式直接精准定位视觉元素，支持对长视频文件（最长10分钟）进行理解和秒级别的事件时刻定位，能理解时间先后和快慢，基于解析和定位能力支持操控OS或Mobile的Agent，关键信息抽取能力和Json格式输出能力强，此版本为72B版本，本系列能力最强的版本。',
displayName: 'Qwen2.5 VL 7B',
id: 'qwen2.5-vl-7b-instruct',
Expand All @@ -412,8 +412,8 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true,
},
contextWindowTokens: 131_072,
description:
'DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力，尤其在数学、代码、自然语言推理等任务上。',
description:
'DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力。在数学、代码、自然语言推理等任务上，性能较高，能力较强。',
displayName: 'DeepSeek R1',
enabled: true,
id: 'deepseek-r1',
Expand All @@ -431,7 +431,7 @@ const qwenChatModels: AIChatModelCard[] = [
functionCall: true,
},
contextWindowTokens: 131_072,
description:
description:
'DeepSeek-V3 为自研 MoE 模型，671B 参数，激活 37B，在 14.8T token 上进行了预训练，在长文本、代码、数学、百科、中文能力上表现优秀。',
displayName: 'DeepSeek V3',
enabled: true,
Expand All @@ -450,8 +450,8 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true,
},
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。',
description:
'DeepSeek-R1-Distill-Qwen-1.5B 是一个基于 Qwen2.5-Math-1.5B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Qwen 1.5B',
id: 'deepseek-r1-distill-qwen-1.5b',
maxOutput: 8192,
Expand All @@ -467,7 +467,7 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true
},
contextWindowTokens: 131_072,
description: "DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。",
description: "DeepSeek-R1-Distill-Qwen-7B 是一个基于 Qwen2.5-Math-7B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。",
displayName: "DeepSeek R1 Distill Qwen 7B",
id: "deepseek-r1-distill-qwen-7b",
maxOutput: 8192,
Expand All @@ -483,9 +483,9 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true
},
contextWindowTokens: 131_072,
description: "DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。",
displayName: "DeepSeek R1 Distill Llama 8B",
id: "deepseek-r1-distill-llama-8b",
description: "DeepSeek-R1-Distill-Qwen-14B 是一个基于 Qwen2.5-14B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。",
displayName: "DeepSeek R1 Distill Qwen 14B",
id: "deepseek-r1-distill-qwen-14b",
maxOutput: 8192,
pricing: {
currency: "CNY",
Expand All @@ -499,9 +499,9 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true
},
contextWindowTokens: 131_072,
description: "DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。",
displayName: "DeepSeek R1 Distill Qwen 14B",
id: "deepseek-r1-distill-qwen-14b",
description: "DeepSeek-R1-Distill-Qwen-32B 是一个基于 Qwen2.5-32B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。",
displayName: "DeepSeek R1 Distill Qwen 32B",
id: "deepseek-r1-distill-qwen-32b",
maxOutput: 8192,
pricing: {
currency: "CNY",
Expand All @@ -515,9 +515,9 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true
},
contextWindowTokens: 131_072,
description: "DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。",
displayName: "DeepSeek R1 Distill Qwen 32B",
id: "deepseek-r1-distill-qwen-32b",
description: "DeepSeek-R1-Distill-Llama-8B 是一个基于 Llama-3.1-8B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。",
displayName: "DeepSeek R1 Distill Llama 8B",
id: "deepseek-r1-distill-llama-8b",
maxOutput: 8192,
pricing: {
currency: "CNY",
Expand All @@ -531,7 +531,7 @@ const qwenChatModels: AIChatModelCard[] = [
reasoning: true
},
contextWindowTokens: 131_072,
description: "DeepSeek-R1-Distill 系列模型通过知识蒸馏技术，将 DeepSeek-R1 生成的样本对 Qwen、Llama 等开源模型进行微调后得到。",
description: "DeepSeek-R1-Distill-Llama-70B 是一个基于 Llama-3.3-70B-Instruct 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。",
displayName: "DeepSeek R1 Distill Llama 70B",
id: "deepseek-r1-distill-llama-70b",
maxOutput: 8192,
Expand Down
90 changes: 84 additions & 6 deletions src/config/modelProviders/qwen.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -294,7 +294,7 @@ const Qwen: ModelProviderCard = {
},
{
contextWindowTokens: 128_000,
description:
description:
'指令跟随、数学、解题、代码整体提升，万物识别能力提升，支持多样格式直接精准定位视觉元素，支持对长视频文件（最长10分钟）进行理解和秒级别的事件时刻定位，能理解时间先后和快慢，基于解析和定位能力支持操控OS或Mobile的Agent，关键信息抽取能力和Json格式输出能力强，此版本为72B版本，本系列能力最强的版本。',
displayName: 'Qwen2.5 VL 72B',
id: 'qwen2.5-vl-72b-instruct',
Expand All @@ -307,9 +307,9 @@ const Qwen: ModelProviderCard = {
vision: true,
},
{
contextWindowTokens: 65_536,
description:
'DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力，尤其在数学、代码、自然语言推理等任务上。',
contextWindowTokens: 131_072,
description:
'DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力。在数学、代码、自然语言推理等任务上，性能较高，能力较强。',
displayName: 'DeepSeek R1',
id: 'deepseek-r1',
pricing: {
Expand All @@ -320,8 +320,8 @@ const Qwen: ModelProviderCard = {
releasedAt: '2025-01-27',
},
{
contextWindowTokens: 65_536,
description:
contextWindowTokens: 131_072,
description:
'DeepSeek-V3 为自研 MoE 模型，671B 参数，激活 37B，在 14.8T token 上进行了预训练，在长文本、代码、数学、百科、中文能力上表现优秀。',
displayName: 'DeepSeek V3',
id: 'deepseek-v3',
Expand All @@ -332,6 +332,84 @@ const Qwen: ModelProviderCard = {
},
releasedAt: '2025-01-27',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Qwen-1.5B 是一个基于 Qwen2.5-Math-1.5B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Qwen 1.5B',
id: 'deepseek-r1-distill-qwen-1.5b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Qwen-7B 是一个基于 Qwen2.5-Math-7B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Qwen 7B',
id: 'deepseek-r1-distill-qwen-7b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Qwen-14B 是一个基于 Qwen2.5-14B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Qwen 14B',
id: 'deepseek-r1-distill-qwen-14b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Qwen-32B 是一个基于 Qwen2.5-32B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Qwen 32B',
id: 'deepseek-r1-distill-qwen-32b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Llama-8B 是一个基于 Llama-3.1-8B 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Llama 8B',
id: 'deepseek-r1-distill-llama-8b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
{
contextWindowTokens: 131_072,
description:
'DeepSeek-R1-Distill-Llama-70B 是一个基于 Llama-3.3-70B-Instruct 的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。',
displayName: 'DeepSeek R1 Distill Llama 70B',
id: 'deepseek-r1-distill-llama-70b',
pricing: {
currency: 'CNY',
input: 0,
output: 0,
},
releasedAt: '2025-02-05',
},
],
checkModel: 'qwen-turbo-latest',
description:
Expand Down