¡Ya disponible! Investigación de Telegram 2025 — los principales insights del año 
Machine learning books and papers
Ir al canal en Telegram
Admin: @Raminmousa ID: @Machine_learn link: https://t.me/Machine_learn
Mostrar más📈 Análisis del canal de Telegram Machine learning books and papers
El canal Machine learning books and papers (@machine_learn) en el segmento lingüístico de Inglés es un actor destacado. Actualmente la comunidad reúne a 24 518 suscriptores, ocupando la posición 8 056 en la categoría Educación y el puesto 13 757 en la región Irán.
📊 Métricas de audiencia y dinámica
Desde su creación el невідомо, el proyecto ha mostrado un crecimiento acelerado, reuniendo a 24 518 suscriptores.
Según los últimos datos del 24 junio, 2026, el canal mantiene una actividad estable. En los últimos 30 días la variación de miembros fue de -165, y en las últimas 24 horas de -3, conservando un alto alcance.
- Estado de verificación: No verificado
- Tasa de interacción (ER): El promedio de interacción de la audiencia es 6.78%. Durante las primeras 24 horas tras publicar, el contenido suele obtener 1.90% de reacciones respecto al total de suscriptores.
- Alcance de las publicaciones: Cada publicación recibe en promedio 1 663 visualizaciones. En el primer día suele acumular 465 visualizaciones.
- Reacciones e interacción: La audiencia responde de forma activa: el promedio de reacciones por publicación es 1.
- Intereses temáticos: El contenido se centra en temas clave como disorder, psy, مقاله, framework, graph.
📝 Descripción y política de contenido
El autor describe el recurso como un espacio para expresar opiniones subjetivas:
“Admin: @Raminmousa
ID: @Machine_learn
link: https://t.me/Machine_learn”
Gracias a la alta frecuencia de actualizaciones (últimos datos recibidos el 25 junio, 2026), el canal mantiene la vigencia y un amplio alcance. La analítica demuestra que la audiencia interactúa activamente con el contenido, lo que lo convierte en un punto de referencia dentro de la categoría Educación.
24 518
Suscriptores
-324 horas
-477 días
-16530 días
Archivo de publicaciones
Artificial Intelligence for Beginners - A Curriculum
📚 Course
@Machine_learn
با عرض سلام نفر سوم از مقاله مروری بالا رو نیاز داریم. با قبولی شرایط کسی نیاز داشت به بنده اطلاع بده.
نشریه مد نظر : Nature
@Raminmousa
Repost from Papers
با عرض سلام در راستاي ادامه تحقيقات مشترك سعي داريم از ١ ام دي ماه روي حوزه ي LLM مدل ها كار كنيم. حدودا ٤ نفر براي كار زير نياز داريم.
BioPars: a pre-trained biomedical large language model for persian biomedical text mining.
١- مراحل اوليه: جمع اوري متن هاي فارسي بيولوژيكي از منابع (...)
٢- پيش پردازش متن ها و تميز كردن متن ها
٣- اموزش ترنسفورمرها ي مورد نظر
٤- استفاده از بردارها ي اموزش داده شده در سه تسك (...)
دوستاني كه مايل به مشاركت هستن مي تونين تا ١ دي بهم اطلاع بدن.
هزينه سرور به ازاي هر ساعت ١.٢ دلار مي باشد. و حدود ٢ هزار ساعت براي اموزش مدل زباني نياز ميباشد. هزينه به ترتيب براي نفرات علاوه بر انجام تسك ها به صورت زير مي باشد.
🔺نفر سوم ٣٠ ميليون
🔹نفر چهارم ٢٥ ميليون
🔺نفر پنجم ٢٠ ميليون
🔹نفر سوم ١٥ ميليون ث
نفرات اول و دوم: رامین موسی و سروش سرابی.
@Raminmousa
@Machine_learn
https://t.me/+SP9l58Ta_zZmYmY0
Primers • Overview of Large Language Models
📖 Link
@Machine_learn
با عرض سلام نفر سوم براي مقاله زير رو خالي داريم.
Title: Alzheimer’s disease (AD) classification
using swin transformer wavelet
and Improved Gray Wolf
Optimization (IGWO)
Abstract: Alzheimer’s disease (AD) is a slow neurological disorder that destroys the thought process, and consciousness, of a human. It directly affects the development of mental ability and neurocognitive functionality. The number of patients with Alzheimer’s disease is increasing day by day, especially in old aged people, who are above 60 years of age, and, gradually, it becomes cause of their death. In this research, our goal is to present ALzSwinTNet for Alzheimer’s classification based on FMRI images. The proposed approach uses wavelet fusion in the swin transformer network to extract features. The igwo and fox optimization approaches were used to find the hyperparameters of the model. ALzSwinTNet was able to achieve an accuracy of 0.98 in 4-class classification and 1 in 2-class classification.
journal: https://www.sciencedirect.com/journal/expert-systems-with-applications
if:7.5
هزینه مشارکت برای نفر سوم ۲۰ تومن می باشد. این هزینه صرف تسویه سرورها خواهد شد.
@Raminmousa
@Machine_learn
https://t.me/+SP9l58Ta_zZmYmY0
📃Graph Neural Networks: A Bibliometric Mapping of the Research Landscape and Applications
📎 Study paper
@Machine_learn
Repost from Papers
با عرض سلام در راستاي ادامه تحقيقات مشترك سعي داريم از ١ ام دي ماه روي حوزه ي LLM مدل ها كار كنيم. حدودا ٥ نفر براي كار زير نياز داريم.
BioPars: a pre-trained biomedical large language model for persian biomedical text mining.
١- مراحل اوليه: جمع اوري متن هاي فارسي بيولوژيكي از منابع (...)
٢- پيش پردازش متن ها و تميز كردن متن ها
٣- اموزش ترنسفورمرها ي مورد نظر
٤- استفاده از بردارها ي اموزش داده شده در سه تسك (...)
دوستاني كه مايل به مشاركت هستن مي تونين تا ١ دي بهم اطلاع بدن.
نكته قابل ذكر اين است كه ما فاندي براي اين كار نداريم و تمامي هزينه ها بين افراد تقسيم ميشه.
@Raminmousa
@Machine_learn
https://t.me/+SP9l58Ta_zZmYmY0
Data Structures and Information Retrieval in Python
📓 link
@Machine_learn
Repost from Papers
با عرض سلام نفر سوم براي مقاله زير رو خالي داريم.
Title: Alzheimer’s disease (AD) classification
using swin transformer wavelet
and Improved Gray Wolf
Optimization (IGWO)
Abstract: Alzheimer’s disease (AD) is a slow neurological disorder that destroys the thought process, and consciousness, of a human. It directly affects the development of mental ability and neurocognitive functionality. The number of patients with Alzheimer’s disease is increasing day by day, especially in old aged people, who are above 60 years of age, and, gradually, it becomes cause of their death. In this research, our goal is to present ALzSwinTNet for Alzheimer’s classification based on FMRI images. The proposed approach uses wavelet fusion in the swin transformer network to extract features. The igwo and fox optimization approaches were used to find the hyperparameters of the model. ALzSwinTNet was able to achieve an accuracy of 0.98 in 4-class classification and 1 in 2-class classification.
journal: https://www.sciencedirect.com/journal/expert-systems-with-applications
if:7.5
هزینه مشارکت برای نفر سوم ۲۰ تومن می باشد. این هزینه صرف تسویه سرورها خواهد شد.
@Raminmousa
@Machine_learn
https://t.me/+SP9l58Ta_zZmYmY0
🌟 INTELLECT-1: Release of the first decentralized learning model.
PRIME Intellect has published INTELLECT-1 ( Instruct + Base ), the first 10 billion parameter language model collaboratively trained in 50 days by 30 participants worldwide.
PRIME Intellect used its own PRIME platform, designed to address the main problems of decentralized learning: network unreliability and dynamic management of computing nodes.
The platform utilized a network of 112 H100 GPUs across 3 continents and achieved a compute utilization rate of 96% under optimal conditions.
The training corpus consisted of 1 trillion public dataset tokens with the following percentage distribution: 55% fineweb-edu, 10% fineweb, 20% Stack V1, 10% dclm-baseline, 5% open-web-math.
▶️ Technical specifications:
🟢 Parameters: 10B;
🟢 Layers: 42;
🟢 Attention Heads: 32;
🟢 Hidden Size: 4096;
🟢 Context Length: 8192;
🟢 Vocabulary Size: 128256.
INTELLECT-1 achieved 37.5% accuracy on the MMLU test and 72.26% on HellaSwag, and outperformed several other open-source models on WinoGrande with a score of 65.82%.
While these figures lag slightly behind today's popular models, the results of the experiment are a critical step toward democratizing AI development and preventing the consolidation of AI capabilities within a few organizations.
▶️ GGUF quantized versions of INTELLECT-1_Instruct in 3-bit (5.46 GB) to 8-bit (10.9 GB) bit depths from the LM Studio community.
▶️ Example of inference on Transformers:
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
torch.set_default_device("cuda")
model = AutoModelForCausalLM.from_pretrained("PrimeIntellect/INTELLECT-1")
tokenizer = AutoTokenizer.from_pretrained("PrimeIntellect/INTELLECT-1")
input_text = "%prompt%"
input_ids = tokenizer.encode(input_text, return_tensors="pt")
output_ids = model.generate(input_ids, max_length=50, num_return_sequences=1)
output_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)
print(output_text)
📌 Licensing: Apache 2.0 License.
🟡 Article
🟡 HF Model Kit
🟡 Set of GGUF versions
🟡 Technical report
🟡 Demo
🖥 GitHub
@Machine_learn
OrientedFormer: An End-to-End Transformer-Based Oriented Object Detector in Remote Sensing Images
Publication date: IEEE Transactions on Geoscience and Remote Sensing 2024
Topic: Object detection
Paper: https://arxiv.org/pdf/2409.19648v1.pdf
GitHub: https://github.com/wokaikaixinxin/OrientedFormer
Description:
In this paper, we propose an end-to-end transformer-based oriented object detector, consisting of three dedicated modules to address these issues. First, Gaussian positional encoding is proposed to encode the angle, position, and size of oriented boxes using Gaussian distributions. Second, Wasserstein self-attention is proposed to introduce geometric relations and facilitate interaction between content and positional queries by utilizing Gaussian Wasserstein distance scores. Third, oriented cross-attention is proposed to align values and positional queries by rotating sampling points around the positional query according to their angles.
@Machine_learn
⚡️ MobileLLM
🟢MobileLLM-125M. 30 Layers, 9 Attention Heads, 3 KV Heads. 576 Token Dimension;
🟢MobileLLM-350M. 32 Layers, 15 Attention Heads, 5 KV Heads. 960 Token Dimension;
🟢MobileLLM-600M. 40 Layers, 18 Attention Heads, 6 KV Heads. 1152 Token Dimension;
🟢MobileLLM-1B. 54 Layers, 20 Attention Heads, 5 KV Heads. 1280 Token Dimension;
🟡Arxiv
🖥GitHub
@Machine_learn
