Skip to content

Commit 3d5e006

Browse files
wenermewenerme
committed
update
1 parent 18c892b commit 3d5e006

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

57 files changed

+1159
-200
lines changed

‎notes/ai/ai-awesome.md‎

Lines changed: 5 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,12 @@ tags:
88
- [LLM Awesome](./llm/llm-awesome.md)
99
- [ML Awesome](./ml/ml-awesome.md)
1010
- [Stable-Diffusion Awesome](./diffusion/diffusion-awesome.md)
11-
- Voice
11+
- [Voice](./voice/voice-awesome.md)
1212
- [jianchang512/clone-voice](https://github.com/jianchang512/clone-voice)
13+
- [FunAudioLLM/CosyVoice](https://github.com/FunAudioLLM/CosyVoice)
14+
- ASR, TTS
15+
- [wenet-e2e/wenet](https://github.com/wenet-e2e/wenet)
16+
- STT
1317
- Search/RAG
1418
- Perplexity
1519
- [ItzCrazyKns/Perplexica](https://github.com/ItzCrazyKns/Perplexica)

‎notes/ai/llm/llm-faq.md‎

Lines changed: 2 additions & 46 deletions
Original file line numberDiff line numberDiff line change
@@ -13,52 +13,8 @@ tags:
1313
- https://huggingface.co/blog/rlhf
1414
- https://github.com/yizhongw/self-instruct
1515
- https://platform.openai.com/docs/model-index-for-researchers
16-
17-
## Pricing
18-
19-
| Model | 1M input/output | notes |
20-
| ------------------------------------: | ---------------: | --------- |
21-
| gpt-4o | $5.00 / $15.00 |
22-
| o1-preview | $15.00 / $60.00 |
23-
| o1-mini | $3.00 / $12.00 |
24-
| gpt-4o-mini | $0.15 / $00.60 |
25-
| Gemini 1.5 Pro 128K | $3.5.00 / $10.50 |
26-
| Gemini 1.5 Flash 128K | $0.075 / $00.30 |
27-
| Gemini 1.5 Flash > 128K | $0.15 / $00.60 |
28-
| Gemini 1.5 {Flash/Pro}-002 | 50% | limits\*2 |
29-
| Claude 3.5 Sonnet 200K | $3.00 / $15.00 |
30-
| Claude 3 Haiku 200K | $0.25 / $01.25 |
31-
| Anthropic Claude 3 Opus | $15.00 / $75.00 |
32-
| Groq Llama 3.1 70B Versatile 128k | $0.59 / $00.79 |
33-
| Groq Whisper V3 Large | $0.111/h |
34-
| DeepInfra Llama-3.1-70B-Instruct 128k | $0.35 / $0.40 |
35-
| DeepInfra Qwen2-72b 32k | $0.35 / $0.40 |
36-
37-
:::tip
38-
39-
- 小模型里 gpt-4o-mini 目前是性价比最好的
40-
- 开源模型的速度可以非常快，能实现 multi agent 这种模式
41-
- Gemini 1.5 Flash 支持 1M context window
42-
- Gemini 1.5 Pro 支持 2M context window
43-
44-
:::
45-
46-
| app | price | quota |
47-
| ------------ | ------------------ | ----------------- |
48-
| ChatGPT Plus | $20 | 4o 80/3h, 4 40/3h |
49-
| ChatGPT Team | $25/年付, $30/月付 | 2\*Plus |
50-
51-
- ChatGPT Plus/Team 限制
52-
- https://help.openai.com/en/articles/6950777-what-is-chatgpt-plus#h_d78bb59065
53-
- https://www.together.ai/pricing
54-
- https://deepinfra.com/pricing
55-
- https://groq.com/pricing/
56-
- https://openai.com/api/pricing/
57-
- https://www.anthropic.com/pricing
58-
- https://fireworks.ai/pricing
59-
- https://www.anyscale.com/pricing
60-
- 参考
61-
- https://www.vellum.ai/blog/llama-3-1-70b-vs-gpt-4o-vs-claude-3-5-sonnet
16+
- Tokenizer
17+
- https://github.com/QwenLM/Qwen/blob/main/tokenization_note_zh.md
6218

6319
## model metrics
6420

‎notes/ai/llm/llm-pricing.md‎

Lines changed: 62 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,62 @@
1+
---
2+
tags:
3+
- VS
4+
- Pricing
5+
---
6+
7+
# Pricing
8+
9+
| Model | 1M input/output | notes |
10+
| :------------------------------------ | :-------------------------------------- | --------- |
11+
| gpt-4o | $5.00 / $15.00 |
12+
| o1 | $15.00 / $60.00 |
13+
| o1-mini | $3.00 / $12.00 |
14+
| gpt-4o-mini | $0.15 / $00.60 |
15+
| OpenAI Realtime API GPT 4o | $5.00 / $20.00, Audio $100.00 / $200.00 |
16+
| OpenAI Whisper | $0.006 / minute |
17+
| OpenAI TTS | $15.00 / 1M chars |
18+
| OpenAI TTS HD | $30.00 / 1M chars |
19+
| Gemini 1.5 Pro 128K | $3.5.00 / $10.50 |
20+
| Gemini 1.5 Flash 128K | $0.075 / $00.30 |
21+
| Gemini 1.5 Flash > 128K | $0.15 / $00.60 |
22+
| Gemini 1.5 {Flash/Pro}-002 | 50% | limits\*2 |
23+
| Claude 3.5 Sonnet 200K | $3.00 / $15.00 |
24+
| Claude 3 Haiku 200K | $0.25 / $01.25 |
25+
| Anthropic Claude 3 Opus | $15.00 / $75.00 |
26+
| Groq Llama 3.1 70B Versatile 128k | $0.59 / $00.79 |
27+
| Groq Whisper V3 Large | $0.111/h |
28+
| DeepInfra Llama-3.1-70B-Instruct 128k | $0.35 / $0.40 |
29+
| DeepInfra Qwen2-72b 32k | $0.35 / $0.40 |
30+
| Aliyun qwen-long 10M | ¥0.50/ ¥2.00 |
31+
| Aliyun qwen-turbo | ¥0.30/ ¥0.60 |
32+
| Aliyun qwen-plus | ¥0.80/ ¥2.00 |
33+
| Aliyun qwen-max | ¥20.00/ ¥60.00 |
34+
35+
:::tip
36+
37+
- 小模型里 gpt-4o-mini 目前是性价比最好的
38+
- 开源模型的速度可以非常快，能实现 multi agent 这种模式
39+
- Gemini 1.5 Flash 支持 1M context window
40+
- Gemini 1.5 Pro 支持 2M context window
41+
42+
:::
43+
44+
| app | price | quota |
45+
| ------------ | ------------------ | ----------------- |
46+
| ChatGPT Plus | $20 | 4o 80/3h, 4 40/3h |
47+
| ChatGPT Team | $25/年付, $30/月付 | 2\*Plus |
48+
49+
- ChatGPT Plus/Team 限制
50+
- https://help.openai.com/en/articles/6950777-what-is-chatgpt-plus#h_d78bb59065
51+
- https://openai.com/api/pricing/
52+
- https://www.together.ai/pricing
53+
- https://deepinfra.com/pricing
54+
- https://groq.com/pricing/
55+
- https://openai.com/api/pricing/
56+
- https://www.anthropic.com/pricing
57+
- https://fireworks.ai/pricing
58+
- https://www.anyscale.com/pricing
59+
- https://openrouter.ai/models?fmt=table
60+
- https://help.aliyun.com/zh/dashscope/developer-reference/tongyi-thousand-questions-metering-and-billing
61+
- 参考
62+
- https://www.vellum.ai/blog/llama-3-1-70b-vs-gpt-4o-vs-claude-3-5-sonnet

‎notes/ai/ml/ml-awesome.md‎

Lines changed: 36 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -171,17 +171,19 @@ tags:
171171
- [facebookresearch/detectron2](https://github.com/facebookresearch/detectron2)
172172
- [open-mmlab/mmdetection](https://github.com/open-mmlab/mmdetection)
173173
- [google-research/big_vision](https://github.com/google-research/big_vision)
174-
- [Yolo](./yolo.md) - You Only Look Once
174+
- [Yolo](./yolo/README.md) - You Only Look Once
175175
- YOLO-NAS - Neural Architecture Search
176176
- [WongKinYiu/yolov7](https://github.com/WongKinYiu/yolov7)
177177
- [YOLOv7 Breakdown](https://blog.roboflow.com/yolov7-breakdown/)
178+
- CLIP
178179
- [lastmile-ai/aiconfig](https://github.com/lastmile-ai/aiconfig)
179180
- MIT, Python
180181
- config-based framework to build generative AI applications
181182
- Dataset
182183
- https://annas-archive.org/llm
183184
- https://www.opendatanetwork.com/
184185
- https://datasetsearch.research.google.com/
186+
- kaggle
185187
- [datumaro](./datumaro.md)
186188
- 数据集管理
187189
- [OpenOrca](https://huggingface.co/datasets/Open-Orca/OpenOrca)
@@ -358,6 +360,39 @@ tags:
358360
- [TheAlgorithms/Python](https://github.com/TheAlgorithms/Python)
359361
- [freeCodeCamp/freeCodeCamp](https://github.com/freeCodeCamp/freeCodeCamp)
360362

363+
## Institute
364+
365+
- OpenAI
366+
- DeepMind
367+
- Microsoft Research
368+
- SAIL - Stanford AI Lab
369+
- Carnegie Mellon University Robotics Institute
370+
- Google AI
371+
- CSAIL - MIT Computer Science and Artificial Intelligence Laboratory
372+
- FAIR - Facebook AI Research
373+
- IBM Research
374+
375+
**国内**
376+
377+
- BAAI - Beijing Academy of Artificial Intelligence - 智源研究院
378+
- [baaivision](https://github.com/baaivision)
379+
- BEG - BAAI General Embedding
380+
- [FlagOpen/FlagEmbedding](https://github.com/FlagOpen/FlagEmbedding)
381+
- Retrieval and Retrieval-augmented LLMs
382+
- https://huggingface.co/BAAI
383+
- https://baai.ac.cn/
384+
- 百度研究院
385+
- https://research.baidu.com/
386+
- 阿里巴巴达摩院
387+
- https://damo.alibaba.com/
388+
- 腾讯 - https://yuanbao.tencent.com/
389+
- 商汤科技
390+
- https://www.sensetime.com/
391+
- 旷视科技
392+
- https://www.megvii.com/
393+
- 云从科技
394+
- https://www.cloudwalk.cn/
395+
361396
## UI/Desktop/GUI/WebUI {#ui}
362397

363398
- omnimodel

‎notes/ai/ml/dataset.md‎ renamed to ‎notes/ai/ml/ml-dataset.md‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -6,6 +6,8 @@ title: Dataset
66

77
- https://roboflow.com/formats
88
- https://github.com/ultralytics/yolov5/blob/master/data/coco128.yaml
9+
- COCO - Common Objects in Context
10+
- by 微软研究院（Microsoft Research） at 2014
911
- coco128
1012
- YOLOv5 Tutorial Dataset
1113
- https://www.kaggle.com/datasets/ultralytics/coco128
@@ -14,3 +16,5 @@ title: Dataset
1416
- [ultralytics/JSON2YOLO](https://github.com/ultralytics/JSON2YOLO)
1517
- Convert JSON annotations into YOLO format
1618
- openlibrary
19+
- 百度数据集 https://aistudio.baidu.com/datasetoverview
20+
- 增值税发票数据集-适配PaddleOCR https://aistudio.baidu.com/datasetdetail/165561

‎notes/ai/ml/ml-glossary.md‎

Lines changed: 32 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -50,6 +50,10 @@ tags:
5050
| MNN |
5151
| TNN |
5252
| NCNN |
53+
| CRNN | Convolutional Recurrent Neural Network | 卷积循环神经网络 |
54+
| DTRB | Deep Transformer Reinforcement Learning | 深度变压器强化学习 |
55+
56+
**Voice**
5357

5458
| abbr. | stand for | cn |
5559
| ----- | -------------------------------------- | ------------------------ |
@@ -67,13 +71,41 @@ tags:
6771
| NMS | Non-Maximum Suppression | 非极大值抑制 |
6872
| IoU | Intersection over Union | 交并比 |
6973
| mAP | Mean Average Precision | 平均精度 |
74+
| SRN | Semantic Reasoning Network | 语义推理网络 |
75+
| STR | scene text recognition | 场景文本识别 |
76+
| SER | Structured Entity Recognition | 结构化实体识别 |
77+
| RE | Relation Extraction | 关系抽取 |
78+
| KIE | Key Information Extraction | 关键信息提取 |
79+
| PSE |
80+
81+
- 文本检测算法
82+
- DB, EAST, SAST, PSE, DB++, FCE
83+
- 文本识别算法
84+
- CRNN, SRN, RARE, NETR, SAR, ViTSTR, ABINet, VisionLAN, SPIN, RobustScanner, SVTR, SVTR_LCNet
85+
- 端到端文本检测算法
86+
- PGNet
87+
88+
**Visual**
89+
90+
| abbr. | stand for | cn |
91+
| ----- | --------------------------------------- | ------------------ |
92+
| MIM | Masked Image Modeling | 掩码图像建模 |
93+
| CLIP | Contrastive Language-Image Pre-training | 对比语言图像预训练 |
94+
| OOB | Oriented Object Detection | 有向物体检测 |
95+
| COCO | Common Objects in Context | 上下文中的通用对象 |
96+
| OKS | Object Keypoint Similarity | 对象关键点相似度 |
7097

7198
| en | cn |
7299
| -------------------- | -------- |
73100
| Contrastive Learning | 对比学习 |
74101
| Inpainting | 局部重绘 |
75102
| Outpainting | 扩展绘制 |
76103

104+
- CLIP - 具有良好的通用性和可扩展性 - modular, reusable, scalable
105+
- MIM - 适合具体的视觉任务 - 如分类、检测、分割
106+
107+
---
108+
77109
- ClassicML
78110
- Regression
79111
- Classification

‎notes/ai/ml/ml-models.md‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
---
2+
tags:
3+
- Models
4+
- Awesome
5+
---
6+
7+
# Models
8+
9+
- [tensorflow/models](https://github.com/tensorflow/models)
10+
- https://www.tensorflow.org/resources/models-datasets
11+
- https://huggingface.co/models
12+
- https://huggingface.co/timm
13+
- ONNX, PyTorch
14+
- https://huggingface.co/spaces/mteb/leaderboard
15+
- https://docs.ultralytics.com/models/
16+
- ONNX
17+
- https://github.com/onnx/models
18+
- PaddlePaddle
19+
- [PaddlePaddle/models](https://github.com/PaddlePaddle/models)
20+
- [PaddlePaddle/PaddleHub](https://github.com/PaddlePaddle/PaddleHub)
21+
- 平台
22+
- https://aistudio.baidu.com/modelsoverview
23+
- https://www.modelscope.cn/
24+
- https://www.modelzoo.co/

‎notes/ai/ml/paddle/paddle-nlp.md‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
---
2+
tags:
3+
- NLP
4+
---
5+
6+
# PaddleNLP
7+
8+
- [PaddlePaddle/PaddleNLP](https://github.com/PaddlePaddle/PaddleNLP)
9+
- Apache-2.0, Python

‎notes/ai/ml/paddle/paddleocr.md‎ renamed to ‎notes/ai/ml/paddle/paddle-ocr.md‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -24,6 +24,8 @@ tags:
2424
- ONNX https://huggingface.co/OleehyO/paddleocrv4.onnx/tree/main
2525
- 缓存目录
2626
- ~/.paddleocr/whl/det/ch/ch_PP-OCRv4_det_infer/ch_PP-OCRv4_det_infer.tar
27+
- 参考
28+
- 模型列表 https://github.com/frotms/PaddleOCR2Pytorch/blob/main/doc/doc_ch/models_list.md
2729

2830
```bash
2931
# 版本 https://pypi.org/project/paddlepaddle/#history

‎notes/ai/ml/paddle/paddle-speech.md‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
---
2+
tags:
3+
- TTS
4+
- ASR
5+
---
6+
7+
# PaddleSpeech
8+
- [PaddlePaddle/PaddleSpeech](https://github.com/PaddlePaddle/PaddleSpeech)

0 commit comments

Comments
 (0)