Channels

Top 100 Trending New

Categories

Crypto Gaming News Entertainment Music Sports Tech Gambling Bots and mini apps Finance Business Education Lifestyle Health Travel Auto Fashion Jobs Dating Deals and shopping

Stickers

Stickers Emoji Themes Wallpapers

Bots Gifts Tools

Currency

$ USD ₽ RUB € EUR ₴ UAH ₸ KZT ₺ TRY zł PLN ₪ ILS AED AED R$ BRL Rp IDR RM MYR ¥ JPY ₩ KRW CN¥ CNY

Theme

Português (Brasil)

Bahasa Indonesia

Беларуская

Українська

Channels Categories Stickers Bots Gifts Tools

Premium Wiki Glossary Legal

Notifications Settings Log in

Bit x Wisdom: post #867 — TG.ME

Bit x Wisdom@bitxwisdom · 48 subscribers Open in TelegramOpen

Bit x WisdomPhoto

.

t.me/bitxwisdom/867 Translate

August 12, 2026 22

More from this channel

engrossed elsewhere for the foreseeable, hence my absence. should the day come, i'll resurface to scribe a line or two.. import torch from transformers import AutoTokenizer, AutoModelForSequenceClassification MODEL_NAME = "cross-encoder/nli-deberta-v3-base" DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu") tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME) model = AutoModelForSequenceClassification.from_pretrained(MODEL_NAME).to(DEVICE) model.eval() FREE_HYPOTHESIS = "The person is free and available." BUSY_HYPOTHESIS = "The person is busy and unavailable." ID_TO_LABEL = { int(index): label.lower() for index, label in model.config.id2label.items() } ENTAILMENT_ID = next( index for index, label in ID_TO_LABEL.items() if "entail" in label ) CONTRADICTION_ID = next( index for index, label in ID_TO_LABEL.items() if "contrad" in label ) ABSENCE_STATE = "BUSY" @torch.inference_mode() def _nli_scores(text, hypothesis): encoded = tokenizer( text, hypothesis, return_tensors="pt", truncation=True, max_length=128 ) encoded = { key: value.to(DEVICE) for key, value in encoded.items() } probabilities = torch.softmax( model(**encoded).logits, dim=-1 )[0] return ( probabilities[ENTAILMENT_ID].item(), probabilities[CONTRADICTION_ID].item() ) def status(text): if not isinstance(text, str) or not text.strip(): return ABSENCE_STATE free_entailment, free_contradiction = _nli_scores( text, FREE_HYPOTHESIS ) busy_entailment, busy_contradiction = _nli_scores( text, BUSY_HYPOTHESIS ) free_score = free_entailment + busy_contradiction busy_score = busy_entailment + free_contradiction return "FREE" if free_score > busy_score else ABSENCE_STATE 23·Aug 29 https://arxiv.org/abs/2608.02859 35·Aug 20 عُمَر جَمیل با ویدیو ۱۹ ساعتِ بعد از ۱ سال برگشت https://youtu.be/XoGvCBRnwLs?si=0SthOvsy0EHTDoXB ربع ساعت اول بحث غیرفنی‌ست(کجا بودم چکار کردم و چی شد و چرا...و معنی یادگیری، چرا باید هنوز یاد گرفت و مسیر و تجارب خودش و....) کد: https://github.com/hkproj/torchfeather Non exhaustive list of papers cited: Attention Is All You Need - https://arxiv.org/abs/... GShard - https://arxiv.org/abs/... Scaling Laws for Fine-Grained Mixture of Experts - https://arxiv.org/abs/... GPipe: https://arxiv.org/abs/... DeepSeek V2: https://arxiv.org/abs/... Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts - https://arxiv.org/abs/... Efficient Large-Scale Language Model Training on GPU Clusters Using Megatron-LM - https://arxiv.org/abs/... Fast Transformer Decoding: One Write-Head is All You Need - https://arxiv.org/abs/... GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints - https://arxiv.org/abs/... Zero Bubble Pipeline Parallelism - https://arxiv.org/abs/... DeepSeek V3 - https://arxiv.org/abs/... Training Compute-Optimal Large Language Models - https://arxiv.org/abs/... Scaling Laws for Neural Language Models - https://arxiv.org/abs/... Chapters 00:00:00 - Introduction 00:16:50 - Model Architecture, Parameters and Training FLOPs 00:51:57 - RoPE from First Principles 01:11:25 - Implementing RoPE and YaRN 01:50:16 - Building the Transformer and Weight Initialization 02:19:12 - Attention, the KV Cache and Arithmetic Intensity 03:02:56 - Coding Multi-head Latent Attention (MLA) 03:15:35 - Block Matrix Multiplication and MLA Internals 03:34:58 - Deriving MLA and Decoupled RoPE 04:10:54 - MLA Weight Absorption 04:32:59 - Autograd and the Mathematics of Distributed Training 05:04:19 - Distributed Computation Graphs and DDP 05:10:47 - Building the Training Loop 06:06:17 - Pipeline Parallelism from First Principles 06:26:52 - Pipeline Schedules: GPipe, 1F1B and Zero Bubble 07:10:29 - Datasets, Tokenization and Data Parallelism 07:42:10 - Coding Pipeline Parallelism 08:37:34 - Device Meshes and Combining PP with DP 09:48:52 - Distributed Communication Collectives 10:49:41 - Implementing Device Meshes, DDP and FSDP 12:13:08 - Tensor Parallelism from First Principles 13:44:57 - Coding Tensor Parallelism 14:47:43 - Context Parallelism and Ring Attention 15:41:22 - Metrics, Optimizers, Schedulers and Checkpointing 16:00:33 - Combining Parallelism in the Training Loop 16:49:58 - Mixture of Experts from First Principles 18:02:52 - Tensor Parallelism for MoE 18:36:24 - Expert Parallelism 19:01:56 - All-to-All Token Dispatch and Combine 19:27:38 - Expert Tensor Parallelism 33·Aug 19 Nothing human makes it out of the near-future. مقاومت بی‌فایده‌ست؛ بهتر است با فرآیند هم‌راستا شوی یا کنار بروی. 29·Aug 18

گایست در مقام صانع امر مطلق تأمّلاتی در بابِ خوانش رابرت براندوم بر معمای قصدیّت، از اسپینوزا تا هگل براندوم در مواجهه با اسپینوزا، کلِّ معماریِ اخلاق را از سطحِ متافیزیک به سطحِ معناشناختی، معرفت‌شناسی و نظریّۀ قصدیّت منتقل می‌کند. براندوم این ادّعای اسپینوزا را به‌رسمیّت می‌شناسد که می‌گوید: یک ایده می‌تواند در ذهنِ من ناکافی باشد در حالی که همان ایده در ذهنِ خدا کافی باشد. مسئلۀ اصلی اما این است که اگر محتوا خاصّیّتِ درونیِ ایده می‌بود، این امر غیرممکن می‌گشت. زیرا یک چیز نمی‌تواند هم‌زمان دارای دو محتوای متفاوت باشد. 📰مطالعۀ متن کامل در وب‌سایت تعمق ✍ #شهام_شریفی Taamoq | تَعَمُّق✅️ 24·Aug 17

An open web layer over public Telegram — search, stats and channel reading right in your browser, no account, no install.

Not affiliated with Telegram FZ-LLC; all data comes from public sources.

Catalog

Top channels Categories Countries Search

Content

Stickers Emoji Gifts Themes Wallpapers

Help

Tools Wiki Glossary Blockchain Safety Premium Advertisers Advertising

© 2026 TG.ME · G.Media network· Updated 09/05/2026 · v0.2.0 · cd00242

Privacy Terms DMCA & complaints Report a violation

Where to next

Hamster Kombat Announcement21.4M subscribers

Blum: All Crypto – One App14.8M subscribers

Proxy MTProto11.9M subscribers

Major Community11.4M subscribers

Telegram Tips11M subscribers

Pavel Durov10.9M subscribers

tapswap community10.6M subscribers

Telegram News9.6M subscribers