Skip to content

Commit 3a1ce41

Browse files
iziangiziang
committed
chore: support deepseek
1 parent 85e3da4 commit 3a1ce41

File tree

3 files changed

+11
-134
lines changed

3 files changed

+11
-134
lines changed

‎aperag/llm/base.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -80,7 +80,7 @@ def match_predictor(model_name, predictor_type, kwargs):
8080
kwargs["model"] = "gpt-4"
8181
from aperag.llm.openai import OpenAIPredictor
8282
return OpenAIPredictor
83-
case "gpt-4-1106-preview" | "gpt-4-vision-preview" | "gpt-4" | "gpt-4-32k" | "gpt-4-0613" | "gpt-4-32k-0613":
83+
case "deepseek-chat" | "gpt-4-1106-preview" | "gpt-4-vision-preview" | "gpt-4" | "gpt-4-32k" | "gpt-4-0613" | "gpt-4-32k-0613":
8484
from aperag.llm.openai import OpenAIPredictor
8585
return OpenAIPredictor
8686
case "azure-openai":

‎deploy/aperag/values.yaml‎

Lines changed: 7 additions & 131 deletions
Original file line numberDiff line numberDiff line change
@@ -176,55 +176,6 @@ django:
176176
memory: "enabled"
177177
# https://help.aliyun.com/zh/dashscope/developer-reference/api-details?spm=a2c4g.11186623.0.i54
178178
context_window: 8096
179-
- name: "chatglm"
180-
label: "ChatGLM"
181-
enabled: "true"
182-
temperature: 0.01
183-
models:
184-
- name: "chatglm-turbo"
185-
label: "ChatGLM Turbo"
186-
enabled: "true"
187-
memory: "enabled"
188-
# context_window not found online
189-
- name: "chatglm-std"
190-
label: "ChatGLM Std"
191-
enabled: "true"
192-
memory: "enabled"
193-
# context_window not found online
194-
- name: "chatglm-lite"
195-
label: "ChatGLM Lite"
196-
enabled: "true"
197-
memory: "enabled"
198-
# context_window not found online
199-
- name: "chatglm-pro"
200-
label: "ChatGLM Pro"
201-
enabled: "true"
202-
memory: "enabled"
203-
# context_window not found online
204-
- name: "chatglm2-6b"
205-
label: "ChatGLM2 6b"
206-
endpoint: "http://llmserver-chatglm2-6b:8000"
207-
enabled: "false"
208-
memory: "disabled"
209-
# https://github.com/THUDM/ChatGLM-6B
210-
context_window: 32384
211-
- name: "baichuan"
212-
label: "BaiChuan"
213-
enabled: "true"
214-
temperature: 0.01
215-
models:
216-
- name: "baichuan-13b"
217-
label: "BaiChuan 13b"
218-
enabled: "false"
219-
memory: "disabled"
220-
# https://www.baichuan-ai.com/home
221-
context_window: 4096
222-
endpoint: "http://llmserver-baichuan-13b:8000"
223-
- name: "baichuan-53b"
224-
label: "BaiChuan 53b"
225-
enabled: "true"
226-
memory: "disabled"
227-
# context_window not found online
228179
- name: "azure-openai"
229180
label: "Azure OpenAI"
230181
enabled: "true"
@@ -242,98 +193,23 @@ django:
242193
enabled: "true"
243194
temperature: 0
244195
models:
245-
- name: "gpt-4-1106-preview"
246-
label: "ChatGPT-4 Turbo"
247-
enabled: "true"
248-
memory: "enabled"
249-
# https://platform.openai.com/docs/models/gpt-4-and-gpt-4-turbo
250-
context_window: 128000
251-
similarity_topk: 10
252196
- name: "gpt-4"
253197
label: "ChatGPT-4"
254198
enabled: "true"
255199
memory: "enabled"
256200
# https://platform.openai.com/docs/models/gpt-4-and-gpt-4-turbo
257201
context_window: 8192
258-
- name: "gpt-4-0613"
259-
label: "ChatGPT-4-0613"
260-
enabled: "true"
261-
memory: "enabled"
262-
# https://platform.openai.com/docs/models/gpt-4-and-gpt-4-turbo
263-
context_window: 8192
264-
- name: "gpt-3.5-turbo-1106"
265-
label: "ChatGPT 3.5 Turbo 1106"
266-
enabled: "true"
267-
memory: "enabled"
268-
# https://platform.openai.com/docs/models/gpt-3-5
269-
context_window: 16385
270-
similarity_topk: 5
271-
- name: "gpt-3.5-turbo"
272-
label: "ChatGPT-3.5 Turbo"
273-
enabled: "true"
274-
memory: "enabled"
275-
# https://platform.openai.com/docs/models/gpt-3-5
276-
context_window: 4096
277-
- name: "gpt-3.5-turbo-16k"
278-
label: "ChatGPT-3.5 Turbo 16k"
279-
enabled: "true"
280-
memory: "enabled"
281-
# https://platform.openai.com/docs/models/gpt-3-5
282-
context_window: 16384
283-
similarity_topk: 5
284-
- name: "wenxinyiyan"
285-
label: "Wen Xin Yi Yan"
202+
- name: "deepseek"
203+
label: "DeepSeek"
286204
enabled: "true"
287205
models:
288-
- name: "ernie-bot-turbo"
289-
label: "Wen Xin Yi Yan"
206+
- name: "deepseek-chat"
207+
label: "DeepSeek Chat"
290208
enabled: "true"
291209
memory: "disabled"
292-
# context_window not found online
293-
- name: "vicuna"
294-
label: "Vicuna"
295-
enabled: "false"
296-
models:
297-
- name: "vicuna-13b"
298-
label: "Vicuna 13b"
299-
enabled: "false"
300-
memory: "disabled"
301-
# https://lmsys.org/blog/2023-03-30-vicuna/
302-
context_window: 2048
303-
endpoint: "http://llmserver-vicuna-13b:8000"
304-
- name: "guanaco"
305-
label: "Guanaco"
306-
enabled: "false"
307-
models:
308-
- name: "guanaco-33b"
309-
label: "Guanaco 33b"
310-
enabled: "false"
311-
memory: "disabled"
312-
# https://llm.extractum.io/model/guanaco-33b-merged,5iVNwg40vN6tl3fT0UJAT1
313-
context_window: 2048
314-
endpoint: "http://llmserver-guanaco-33b:8000"
315-
- name: "falcon"
316-
label: "Falcon"
317-
enabled: "false"
318-
models:
319-
- name: "falcon-40b"
320-
label: "Falcon 40b"
321-
enabled: "false"
322-
memory: "disabled"
323-
# https://codingscape.com/blog/most-powerful-llms-large-language-models-in-2023
324-
context_window: 2048
325-
endpoint: "http://llmserver-falcon-40b:8000"
326-
- name: "gorilla"
327-
label: "Gorilla"
328-
enabled: "false"
329-
models:
330-
- name: "gorilla-7b"
331-
label: "Gorilla 7b"
332-
enabled: "false"
333-
memory: "disabled"
334-
# https://llm.extractum.io/model/gorilla-llm/gorilla-7b-hf-delta-v0,628M0wYI5qNZeBLak6s82K
335-
context_window: 2048
336-
endpoint: "http://llmserver-gurilla-7b:8000"
210+
free_tier: true
211+
context_window: 128000
212+
similarity_topk: 10
337213

338214
celery-worker:
339215
replicaCount: 1

‎envs/.env.template‎

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
LLM_MODEL=vicuna-13b
2-
MODEL_FAMILIES=[{"name":"qianwen","label":"QianWen","enabled":"true","temperature":0.01,"models":[{"name":"qwen-turbo","label":"QianWen Turbo","enabled":"true","memory":"enabled"},{"name":"qwen-plus","label":"QianWen Plus","enabled":"true","memory":"enabled"},{"name":"qwen-max","label":"QianWen Max","enabled":"true","memory":"enabled"}]},{"name":"chatglm","label":"ChatGLM","enabled":"true","temperature":0.01,"models":[{"name":"chatglm-turbo","label":"ChatGLM Turbo","enabled":"true","memory":"enabled"},{"name":"chatglm-std","label":"ChatGLM Std","enabled":"true","memory":"enabled"},{"name":"chatglm-lite","label":"ChatGLM Lite","enabled":"true","memory":"enabled"},{"name":"chatglm-pro","label":"ChatGLM Pro","enabled":"true","memory":"enabled"},{"name":"chatglm2-6b","label":"ChatGLM2 6b","endpoint":"http://llmserver-chatglm2-6b:8000","enabled":"false","memory":"disabled"}]},{"name":"baichuan","label":"BaiChuan","enabled":"true","temperature":0.01,"models":[{"name":"baichuan-13b","label":"BaiChuan 13b","enabled":"false","memory":"disabled","endpoint":"http://llmserver-baichuan-13b:8000"},{"name":"baichuan-53b","label":"BaiChuan 53b","enabled":"true","memory":"disabled"}]},{"name":"azure-openai","label":"Azure OpenAI","enabled":"true","temperature":0,"models":[{"name":"azure-openai","label":"Azure OpenAI","enabled":"true","memory":"enabled"}]},{"name":"chatgpt","label":"ChatGPT","enabled":"true","temperature":0,"models":[{"name":"gpt-4-1106-preview","label":"ChatGPT-4 Turbo","enabled":"true","memory":"enabled","context_window":128000,"similarity_topk":10},{"name":"gpt-4","label":"ChatGPT-4","enabled":"true","memory":"enabled","context_window":8192},{"name":"gpt-4-0613","label":"ChatGPT-4-0613","enabled":"true","memory":"enabled","context_window":8192},{"name":"gpt-3.5-turbo-1106","label":"ChatGPT 3.5 Turbo 1106","enabled":"true","memory":"enabled","context_window":16385,"similarity_topk":5},{"name":"gpt-3.5-turbo","label":"ChatGPT-3.5 Turbo","enabled":"true","memory":"enabled","context_window":4096},{"name":"gpt-3.5-turbo-16k","label":"ChatGPT-3.5 Turbo 16k","enabled":"true","memory":"enabled","context_window":16384,"similarity_topk":5}]},{"name":"wenxinyiyan","label":"Wen Xin Yi Yan","enabled":"true","models":[{"name":"ernie-bot-turbo","label":"Wen Xin Yi Yan","enabled":"true","memory":"disabled"}]},{"name":"vicuna","label":"Vicuna","enabled":"false","models":[{"name":"vicuna-13b","label":"Vicuna 13b","enabled":"false","memory":"disabled","endpoint":"http://llmserver-vicuna-13b:8000"}]},{"name":"guanaco","label":"Guanaco","enabled":"false","models":[{"name":"guanaco-33b","label":"Guanaco 33b","enabled":"false","memory":"disabled","endpoint":"http://llmserver-guanaco-33b:8000"}]},{"name":"falcon","label":"Falcon","enabled":"false","models":[{"name":"falcon-40b","label":"Falcon 40b","enabled":"false","memory":"disabled","endpoint":"http://llmserver-falcon-40b:8000"}]},{"name":"gorilla","label":"Gorilla","enabled":"false","models":[{"name":"gorilla-7b","label":"Gorilla 7b","enabled":"false","memory":"disabled","endpoint":"http://llmserver-gurilla-7b:8000"}]}]
2+
MODEL_FAMILIES=[{"name":"qianwen","label":"QianWen","enabled":"true","temperature":0.01,"models":[{"name":"qwen-turbo","label":"QianWen Turbo","enabled":"true","memory":"enabled","context_window":8096},{"name":"qwen-plus","label":"QianWen Plus","enabled":"true","memory":"enabled","context_window":8096},{"name":"qwen-max","label":"QianWen Max","enabled":"true","memory":"enabled","context_window":8096}]},{"name":"azure-openai","label":"Azure OpenAI","enabled":"true","temperature":0,"models":[{"name":"azure-openai","label":"Azure OpenAI","enabled":"true","memory":"enabled","context_window":4096,"free_tier":true}]},{"name":"chatgpt","label":"ChatGPT","enabled":"true","temperature":0,"models":[{"name":"gpt-4","label":"ChatGPT-4","enabled":"true","memory":"enabled","context_window":8192}]},{"name":"deepseek","label":"DeepSeek","enabled":"true","models":[{"name":"deepseek-chat","label":"DeepSeek Chat","enabled":"true","memory":"disabled","free_tier":true,"context_window":128000,"similarity_topk":10}]}]
33
LIMIT_MODEL_CONCURRENCY=5
44
MAX_POSITION_EMBEDDINGS=4096
55
QUANTIZE_QLORA=True
@@ -57,7 +57,6 @@ EMBEDDING_SERVICE_URL=http://localhost:9997
5757
EMBEDDING_SERVICE_MODEL=bge-large-zh-v1.5
5858
EMBEDDING_SERVICE_TOKEN=
5959
EMBEDDING_SERVICE_MODEL_UID=
60-
OPENAI_API_KEY=
6160

6261

6362
CHAT_CONSUMER_IMPLEMENTATION=document-qa
@@ -81,3 +80,5 @@ DEFAULT_ENCODING_MODEL=cl100k_base
8180
DATABASE_URL="postgres://postgres:postgres@127.0.0.1:5432/postgres"
8281

8382
#OPENAI_API_PROXY='{"https": "socks5h://127.0.0.1:1080"}'
83+
OPENAI_API_BASE=
84+
OPENAI_API_KEY=

0 commit comments

Comments
 (0)