Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

Motel Owner, Dylan Patel, talks about GPT6's architecture which will continue the trend of increasing MoE sparsity. This will require even wider expert parallelism for MoE dispatch & MoE combine collectives during decode phase.

SemiAnalysis

109,487 subscribers

37,640 görüntüleme • 5 ay önce •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

llmfit. Useful tool that probes hardware and tells you exactly which LLMs will actually run. - handles MoE expert offloading, picks the best quantization for your RAM, estimates tokens/sec before you even pull the weights. Essential for local dev.

llmfit. Useful tool that probes hardware and tells you exactly which LLMs will actually run. - handles MoE expert offloading, picks the best quantization for your RAM, estimates tokens/sec before you even pull the weights. Essential for local dev.

Wildminder

248,279 görüntüleme • 6 ay önce

On large scale frontier MoE models, we are seeing GB200 NVL72 FP4 is 10x more power efficient compared to H200 FP8 single node at certain interactivity (tok/s/user). We are excited to implement disagg prefill/multinode wide expert parallelism on h100/h200 to see what is the power efficient gains between GB200 & H200 once we have that implemented. The power efficiency will probably be reduced to 4-6x once we have that implemented on hopper, which is still pretty good! visit inferencemax dot ai for the latest nuanced results.

On large scale frontier MoE models, we are seeing GB200 NVL72 FP4 is 10x more power efficient compared to H200 FP8 single node at certain interactivity (tok/s/user). We are excited to implement disagg prefill/multinode wide expert parallelism on h100/h200 to see what is the power efficient gains between GB200 & H200 once we have that implemented. The power efficiency will probably be reduced to 4-6x once we have that implemented on hopper, which is still pretty good! visit inferencemax dot ai for the latest nuanced results.

SemiAnalysis

56,872 görüntüleme • 10 ay önce

New short course with Mistral AI ! Mistral's open-source Mixtral 8x7B model uses a "mixture of experts" (MoE) architecture. Unlike a standard transformer, an MoE model has multiple expert feed-forward networks (8 in this case), with a gating network selecting two experts at inference time. This enables MoE to match the performance of a large model but faster inference. Mixtral 8x7B has 46.7B parameters but activates only 12.9B at inference to predict the next token. In Getting Started with Mistral, you’ll learn from Mistral’s Sophia Yang, Ph.D. to: - Explore Mistral's open-source models (Mistral 7B, Mixtral 8x7B) and commercial models via API calls and Mistral AI's Le Chat website - Implement JSON mode to generate structured outputs to integrate directly into larger software systems. - Use function calling for Tool Use, such as calling custom Python code that queries tabular data - Ground your LLM's response with external knowledge sources using RAG - Build a Mistral-powered chat interface that can reference external documents This course will help deepen your prompt engineering skills. Please sign up here:

New short course with Mistral AI ! Mistral's open-source Mixtral 8x7B model uses a "mixture of experts" (MoE) architecture. Unlike a standard transformer, an MoE model has multiple expert feed-forward networks (8 in this case), with a gating network selecting two experts at inference time. This enables MoE to match the performance of a large model but faster inference. Mixtral 8x7B has 46.7B parameters but activates only 12.9B at inference to predict the next token. In Getting Started with Mistral, you’ll learn from Mistral’s Sophia Yang, Ph.D. to: - Explore Mistral's open-source models (Mistral 7B, Mixtral 8x7B) and commercial models via API calls and Mistral AI's Le Chat website - Implement JSON mode to generate structured outputs to integrate directly into larger software systems. - Use function calling for Tool Use, such as calling custom Python code that queries tabular data - Ground your LLM's response with external knowledge sources using RAG - Build a Mistral-powered chat interface that can reference external documents This course will help deepen your prompt engineering skills. Please sign up here:

Andrew Ng

387,172 görüntüleme • 2 yıl önce

After Alberta issued an emergency alert about the possibility of rolling blackouts because of high power demands on their grid during the extreme cold, Saskatchewan fired up their coal and gas plants to provide Alberta with an extra 153 MW of electricity through the night. Sask Scott Moe posted, “That power will be coming from natural gas and coal-fired plants, the ones the Trudeau government is telling us to shut down (which we won’t).” #saskatchewan | #alberta | #canada | #carbontax | #polarvortex | #winter | #trudeau

After Alberta issued an emergency alert about the possibility of rolling blackouts because of high power demands on their grid during the extreme cold, Saskatchewan fired up their coal and gas plants to provide Alberta with an extra 153 MW of electricity through the night. Sask Scott Moe posted, “That power will be coming from natural gas and coal-fired plants, the ones the Trudeau government is telling us to shut down (which we won’t).” #saskatchewan | #alberta | #canada | #carbontax | #polarvortex | #winter | #trudeau

Just Bins

159,769 görüntüleme • 2 yıl önce

Soofi Consortium Releases Soofi S 30B-A3B: An Open 31.6B Model for German and English Hitting 79.1 German Aggregate With Only 3.2B Active Parameters. Here's how it works. 👇 1. Sparsity in two places at once 52 layers: 23 Mamba-2, 23 granular MoE, 6 Grouped-Query Attention. The MoE router picks 6 of 128 experts per token, plus 2 shared. Mamba-2 carries the sequence mixing with a fixed-size recurrent state, so 46 of 52 layers keep no KV cache at all. → 3.2B of 31.6B parameters active per token 2. Reference architecture on purpose No bespoke backbone. It adopts NVIDIA's Nemotron 3 Nano design without modification — for day-one vLLM kernels, for serving efficiency, and for scientific control. That last one is the real move: Nemotron becomes an architecture-identical baseline, so the data recipe is the only variable left. 3. German as the deliberate variable Three-phase Warmup–Stable–Decay curriculum. Phase 1 is breadth at a 1e-3 plateau, Phase 2 concentrates high-quality data as the LR decays, Phase 3 stretches context to 1M tokens. → ~26.68T consumed tokens → German 7.2% → 15.32% of the mixture, vs ~5% for all non-English in the Nemotron reference → +4.2 German aggregate, +1.8 English, +6.7 held-out English over Nemotron 4. Where the architecture pays: memory bandwidth Every decoded token re-reads the weights and, for a Transformer, the attention cache of every sequence in the batch. Six KV layers instead of 52 keeps that per-sequence state small. Measured on one B200, TP=1, vLLM latency-subtraction. → 8–9× aggregate decode TPS/GPU vs dense 14–24B models at 40K context, batch 32 → decode stays flat from 4K to 256K 5. The numbers (base model, lm-evaluation-harness, 16 open baselines) → 70.1 English aggregate, +2.8 over Olmo 3 32B → 79.1 German aggregate, +6.3 over Apertus 70B → 73.8 HumanEval, 84.2 MBPP-DE, 88.8 GLP-DE, 61.2 INCLUDE-DE Full analysis: Paper: Technical details:

Soofi Consortium Releases Soofi S 30B-A3B: An Open 31.6B Model for German and English Hitting 79.1 German Aggregate With Only 3.2B Active Parameters. Here's how it works. 👇 1. Sparsity in two places at once 52 layers: 23 Mamba-2, 23 granular MoE, 6 Grouped-Query Attention. The MoE router picks 6 of 128 experts per token, plus 2 shared. Mamba-2 carries the sequence mixing with a fixed-size recurrent state, so 46 of 52 layers keep no KV cache at all. → 3.2B of 31.6B parameters active per token 2. Reference architecture on purpose No bespoke backbone. It adopts NVIDIA's Nemotron 3 Nano design without modification — for day-one vLLM kernels, for serving efficiency, and for scientific control. That last one is the real move: Nemotron becomes an architecture-identical baseline, so the data recipe is the only variable left. 3. German as the deliberate variable Three-phase Warmup–Stable–Decay curriculum. Phase 1 is breadth at a 1e-3 plateau, Phase 2 concentrates high-quality data as the LR decays, Phase 3 stretches context to 1M tokens. → ~26.68T consumed tokens → German 7.2% → 15.32% of the mixture, vs ~5% for all non-English in the Nemotron reference → +4.2 German aggregate, +1.8 English, +6.7 held-out English over Nemotron 4. Where the architecture pays: memory bandwidth Every decoded token re-reads the weights and, for a Transformer, the attention cache of every sequence in the batch. Six KV layers instead of 52 keeps that per-sequence state small. Measured on one B200, TP=1, vLLM latency-subtraction. → 8–9× aggregate decode TPS/GPU vs dense 14–24B models at 40K context, batch 32 → decode stays flat from 4K to 256K 5. The numbers (base model, lm-evaluation-harness, 16 open baselines) → 70.1 English aggregate, +2.8 over Olmo 3 32B → 79.1 German aggregate, +6.3 over Apertus 70B → 73.8 HumanEval, 84.2 MBPP-DE, 88.8 GLP-DE, 61.2 INCLUDE-DE Full analysis: Paper: Technical details:

Marktechpost AI

65,592 görüntüleme • 1 ay önce

I trained a 100 million parameter DeepSeek V3 LLM from scratch Here's what you need to know. Previously I trained traditional GPT-2 architecture which has become obsolete with recent LLM advancements. Most recent models like Llama, Mistral, DeepSeek, and GPT-4 use latest architectures. ✦ Model Configuration of my SLM DeepSeek V3 - Parameters: 109,032,032 - Embedding Dimension: 512 - Layers: 8 - Heads: 8 - Experts (MoE): 8 - Experts per token: 2 ✦ DeepSeek brings major architectural changes: - Multi Head Latent Attention - Mixture of Experts - RMS Norm - Multi Token Prediction ✦ Dataset Challenge - TinyStories is great for learning SLMs. I trained GPT-2 on it previously with good results. - But I needed a more challenging dataset. - If I use TinyStories again on DeepSeek, how would I know MHLA, MoE or MTP works better than old architecture? - The old architecture can handle it, so new DeepSeek would too without utilizing latest advancements. That's why I moved to FineWeb-Edu dataset Thanks Yuvraj Singh for the suggestion for this dataset ✦ Training Journey - Rented A100 PCIe GPU and trained the model. - Did test runs. During final run, model was 65% trained but stopped due to glitch after 4 hours. - Fixed all edge cases and ran training again with increased config parameters. - Final training: 7 hours, 20,000 epochs 𝐓𝐨𝐭𝐚𝐥 𝐆𝐏𝐔 𝐜𝐨𝐬𝐭: $17 - $9.53 for main 7-hour run - $7.42 for experiments and demos ✦ Reflection Amazing long project that taught me latest architectural advancements. I'll reimplement and revisit after a few weeks because there's too much complexity, mostly in Multi Head Latent Attention part. Need to make concepts stronger. Code Final trained Model Dataset Resources Huge shoutout to Raj Dandekar again for creating one of the most detailed video series about DeepSeek - this was my primary resource for the implementation. Playlist Blogs by Maarten Grootendorst These are excellent visual blogs to understand MoE in detail. Thanks Maarten for your amazing contributions to the community through your books and blogs Blogs on MoE Implemention of MoE from scratch by @aviTwit3 One of the most detailed blogs on implementing Mixture of Experts. Thanks Avinash for this blog - it helped me understand Mixture of Experts much better. If you're someone in the 𝐌𝐋 & 𝐋𝐋𝐌 space, would love to 𝐜𝐨𝐧𝐧𝐞𝐜𝐭 and discuss this field in general, so give a follow up for that.

I trained a 100 million parameter DeepSeek V3 LLM from scratch Here's what you need to know. Previously I trained traditional GPT-2 architecture which has become obsolete with recent LLM advancements. Most recent models like Llama, Mistral, DeepSeek, and GPT-4 use latest architectures. ✦ Model Configuration of my SLM DeepSeek V3 - Parameters: 109,032,032 - Embedding Dimension: 512 - Layers: 8 - Heads: 8 - Experts (MoE): 8 - Experts per token: 2 ✦ DeepSeek brings major architectural changes: - Multi Head Latent Attention - Mixture of Experts - RMS Norm - Multi Token Prediction ✦ Dataset Challenge - TinyStories is great for learning SLMs. I trained GPT-2 on it previously with good results. - But I needed a more challenging dataset. - If I use TinyStories again on DeepSeek, how would I know MHLA, MoE or MTP works better than old architecture? - The old architecture can handle it, so new DeepSeek would too without utilizing latest advancements. That's why I moved to FineWeb-Edu dataset Thanks Yuvraj Singh for the suggestion for this dataset ✦ Training Journey - Rented A100 PCIe GPU and trained the model. - Did test runs. During final run, model was 65% trained but stopped due to glitch after 4 hours. - Fixed all edge cases and ran training again with increased config parameters. - Final training: 7 hours, 20,000 epochs 𝐓𝐨𝐭𝐚𝐥 𝐆𝐏𝐔 𝐜𝐨𝐬𝐭: $17 - $9.53 for main 7-hour run - $7.42 for experiments and demos ✦ Reflection Amazing long project that taught me latest architectural advancements. I'll reimplement and revisit after a few weeks because there's too much complexity, mostly in Multi Head Latent Attention part. Need to make concepts stronger. Code Final trained Model Dataset Resources Huge shoutout to Raj Dandekar again for creating one of the most detailed video series about DeepSeek - this was my primary resource for the implementation. Playlist Blogs by Maarten Grootendorst These are excellent visual blogs to understand MoE in detail. Thanks Maarten for your amazing contributions to the community through your books and blogs Blogs on MoE Implemention of MoE from scratch by @aviTwit3 One of the most detailed blogs on implementing Mixture of Experts. Thanks Avinash for this blog - it helped me understand Mixture of Experts much better. If you're someone in the 𝐌𝐋 & 𝐋𝐋𝐌 space, would love to 𝐜𝐨𝐧𝐧𝐞𝐜𝐭 and discuss this field in general, so give a follow up for that.

Mayank Pratap Singh

48,120 görüntüleme • 1 yıl önce

Dylan Patel: “CPUs are sold out.” $INTC earnings and conference call confirmed the looming CPU shortage. There are three main drivers: - Agentic systems, as CPUs do the planning and orchestration. - Increasing complexity of and demand for reinforcement learning environments that run on CPUs. - Increasing deployment of AI instilled apps. Models run on GPUs but apps themselves run on CPUs. Demand for all these will grow secularly as AI systems will only get more agentic, RL environments will only get more complex, and deployment will only accelerate. As a result, GPU-to-CPU ratio in AI workloads will move from 1:8 to 1:2 or maybe even 1:1 in the future. CPU supercycle is about to start. $AMD $INTC $ARM

Dylan Patel: “CPUs are sold out.” $INTC earnings and conference call confirmed the looming CPU shortage. There are three main drivers: - Agentic systems, as CPUs do the planning and orchestration. - Increasing complexity of and demand for reinforcement learning environments that run on CPUs. - Increasing deployment of AI instilled apps. Models run on GPUs but apps themselves run on CPUs. Demand for all these will grow secularly as AI systems will only get more agentic, RL environments will only get more complex, and deployment will only accelerate. As a result, GPU-to-CPU ratio in AI workloads will move from 1:8 to 1:2 or maybe even 1:1 in the future. CPU supercycle is about to start. $AMD $INTC $ARM

Oguz Erkan

93,419 görüntüleme • 4 ay önce

Prosecutors say OTF Jam will allegedly testify that the $50k Lil Durk gave him (that he rapped about in Back Again) was in his mind, a way to get him (Jam) to do things for Durk. Jam will also allegedly testify that when they filming a music video, Durk and some other OTF members took him into a bathroom and told him to chill on the murders cuz it would get the Feds on them. Trap says that it sounded like the prosecutors were saying Jam will be testifying on other murders not even related to Lul Pab’s. Jam will allegedly be testifying that he was recruited for the Lul Pab murder by Boonie Moe on behalf of Durk

Prosecutors say OTF Jam will allegedly testify that the $50k Lil Durk gave him (that he rapped about in Back Again) was in his mind, a way to get him (Jam) to do things for Durk. Jam will also allegedly testify that when they filming a music video, Durk and some other OTF members took him into a bathroom and told him to chill on the murders cuz it would get the Feds on them. Trap says that it sounded like the prosecutors were saying Jam will be testifying on other murders not even related to Lul Pab’s. Jam will allegedly be testifying that he was recruited for the Lul Pab murder by Boonie Moe on behalf of Durk

Akademiks TV

216,534 görüntüleme • 14 gün önce

‘We will continue to arm Ukraine and hope for a ceasefire’ – Merz on Europe’s plans to finally break with common sense. ‘We do not rule out any conceivable possibility of reaching a diplomatic solution. We understand that it will not be possible to achieve it in the shortest possible time. Therefore, we will continue to support Ukraine, including militarily. At the same time, we hope that there will be a chance for a ceasefire, and then for peace talks. But once again: this is a process that is just beginning and which may last many more weeks, perhaps even months.’ - SK

‘We will continue to arm Ukraine and hope for a ceasefire’ – Merz on Europe’s plans to finally break with common sense. ‘We do not rule out any conceivable possibility of reaching a diplomatic solution. We understand that it will not be possible to achieve it in the shortest possible time. Therefore, we will continue to support Ukraine, including militarily. At the same time, we hope that there will be a chance for a ceasefire, and then for peace talks. But once again: this is a process that is just beginning and which may last many more weeks, perhaps even months.’ - SK

Zlatti71

25,158 görüntüleme • 1 yıl önce

The VRAM barrier is officially dead. I just ran Qwen 3.8 Flash Next (MoE) 125B A6B with a 250,000 context window on a single 24GB RTX 4090. 21 tokens/sec decode. 364 t/s prefill. no mtp. no dflash. no kv cache quantization! We are running datacenter models on consumer hardware. Tested on Ubuntu 22 | CUDA 13.0 | PCIe 4.0 x16 | 110 GB DDR4 System RAM with a continuous 28k prompt across all runs. ### The Benchmarks & Scaling # 1. Hybrid Offload (-ncmoe 40 @ 80k Context) Offloaded 40 expert layers to the GPU, pushing VRAM to the ceiling. ./build/bin/llama-server -m Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf -c 80000 --port 8080 -v --fit off -b 4096 -ub 4096 -ncmoe 40 Prefill: 383.85 t/s | Decode: 22.52 t/s Footprint: 23.85 GB VRAM | 97 GB RAM # 2. Full CPU MoE Offload (-cmoe @ 80k Context) Pinned all 512 expert layers to DDR4 RAM (-cmoe), keeping attention on the 4090. llama.cpp flags: (Same as above, replace -ncmoe 40 with -cmoe) Prefill: 355.72 t/s | Decode: 20.84 t/s Footprint: 11.66 GB VRAM (12GB+ VRAM freed up!) | 110 GB RAM # 3. The 180,000 Context Run Prefill: 357.75 t/s | Decode: 20.98 t/s | VRAM: 15.6 GB | RAM: 110 GB # 4. The 250,000 Context Absolute Ceiling ./build/bin/llama-server -m Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf -c 250000 --port 8080 -v --fit off -b 4096 -ub 4096 -cmoe Prefill: 364.29 t/s | Decode: 20.97 t/s Footprint: 18.3 GB VRAM (Still ~5.7 GB of VRAM headroom!) | 110 GB RAM ### Key Insights: -b 4096 -ub 4096: doubles the prompt ingestion from ~150 to 364+ t/s. -cmoe Free Lunch: Shifting expert layers to DDR4 RAM slashes VRAM from 24GB to 11.6GB with virtually zero decode penalty (22.5 -> 20.9 t/s), enabling the 250k context ceiling. Qwen 3.8 Flash-Next (UD-Q4_K_XL) is a massive 111.4 GB model split across 4 shards. To run this architecture, you must build from the experimental PR branch (#27742) by Daniel Han: git clone && cd llama.cpp git fetch origin pull/27742/head:qwen-next && git checkout qwen-next cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=native -DBUILD_SHARED_LIBS=OFF cmake --build build --config Release -j $(nproc) --target llama-server A single 4090 paired with 100 GB of cheap DDR4 RAM will comfortably serve production grade 125B inference. While Qwen 3.8 27B (dense) still holds the crown for single 3090/4090 rigs, Flash Next proves 125B hybrid models are officially viable on consumer hardware. Hugging Face GGUF link and complete performance telemetry graphs are dropped in the replies below. GLM 5.3 Flash VS Qwen 3.8 Flash Next, which one takes the open weights crown this week?

The VRAM barrier is officially dead. I just ran Qwen 3.8 Flash Next (MoE) 125B A6B with a 250,000 context window on a single 24GB RTX 4090. 21 tokens/sec decode. 364 t/s prefill. no mtp. no dflash. no kv cache quantization! We are running datacenter models on consumer hardware. Tested on Ubuntu 22 | CUDA 13.0 | PCIe 4.0 x16 | 110 GB DDR4 System RAM with a continuous 28k prompt across all runs. ### The Benchmarks & Scaling # 1. Hybrid Offload (-ncmoe 40 @ 80k Context) Offloaded 40 expert layers to the GPU, pushing VRAM to the ceiling. ./build/bin/llama-server -m Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf -c 80000 --port 8080 -v --fit off -b 4096 -ub 4096 -ncmoe 40 Prefill: 383.85 t/s | Decode: 22.52 t/s Footprint: 23.85 GB VRAM | 97 GB RAM # 2. Full CPU MoE Offload (-cmoe @ 80k Context) Pinned all 512 expert layers to DDR4 RAM (-cmoe), keeping attention on the 4090. llama.cpp flags: (Same as above, replace -ncmoe 40 with -cmoe) Prefill: 355.72 t/s | Decode: 20.84 t/s Footprint: 11.66 GB VRAM (12GB+ VRAM freed up!) | 110 GB RAM # 3. The 180,000 Context Run Prefill: 357.75 t/s | Decode: 20.98 t/s | VRAM: 15.6 GB | RAM: 110 GB # 4. The 250,000 Context Absolute Ceiling ./build/bin/llama-server -m Qwen3.8-Flash-Next-UD-Q4_K_XL-00001-of-00004.gguf -c 250000 --port 8080 -v --fit off -b 4096 -ub 4096 -cmoe Prefill: 364.29 t/s | Decode: 20.97 t/s Footprint: 18.3 GB VRAM (Still ~5.7 GB of VRAM headroom!) | 110 GB RAM ### Key Insights: -b 4096 -ub 4096: doubles the prompt ingestion from ~150 to 364+ t/s. -cmoe Free Lunch: Shifting expert layers to DDR4 RAM slashes VRAM from 24GB to 11.6GB with virtually zero decode penalty (22.5 -> 20.9 t/s), enabling the 250k context ceiling. Qwen 3.8 Flash-Next (UD-Q4_K_XL) is a massive 111.4 GB model split across 4 shards. To run this architecture, you must build from the experimental PR branch (#27742) by Daniel Han: git clone && cd llama.cpp git fetch origin pull/27742/head:qwen-next && git checkout qwen-next cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=native -DBUILD_SHARED_LIBS=OFF cmake --build build --config Release -j $(nproc) --target llama-server A single 4090 paired with 100 GB of cheap DDR4 RAM will comfortably serve production grade 125B inference. While Qwen 3.8 27B (dense) still holds the crown for single 3090/4090 rigs, Flash Next proves 125B hybrid models are officially viable on consumer hardware. Hugging Face GGUF link and complete performance telemetry graphs are dropped in the replies below. GLM 5.3 Flash VS Qwen 3.8 Flash Next, which one takes the open weights crown this week?

Alok

990,151 görüntüleme • 6 gün önce

"Every time there are talks scheduled between Iran and the United States, it's open season for hostage taking in Iran." Siamak Namazi, Morad Tahbaz, and Emad Shargi are speaking out about their concerns with US, Iran talks set to continue-- "now we are entering a new phase of discussions, and the three of us fear that this will cause these elements within the regime, within the IRGC to go out looking to fill up their cells"

"Every time there are talks scheduled between Iran and the United States, it's open season for hostage taking in Iran." Siamak Namazi, Morad Tahbaz, and Emad Shargi are speaking out about their concerns with US, Iran talks set to continue-- "now we are entering a new phase of discussions, and the three of us fear that this will cause these elements within the regime, within the IRGC to go out looking to fill up their cells"

Christiane Amanpour

31,182 görüntüleme • 1 yıl önce

One of the hardest things for Conservatives and Pierre Poilievre to overcome will be the fact that Mark Carney gets praised by all of Canada's Premiers and they are pretty happy with the work he is doing. Saskatchewan Premier Scott Moe: "We have seen Prime Minister Mark Carney greatly elevate Canada's credibility on the global stage. If we keep looking in the rear view mirror I think we will keep increasingly frustrated and as you look out of the windshield we are seeing some quiet different from this Prime Minister." The lost decade of Justin Trudeau Liberals have created a contrast...but which of these disastrous policies have been removed so far?

One of the hardest things for Conservatives and Pierre Poilievre to overcome will be the fact that Mark Carney gets praised by all of Canada's Premiers and they are pretty happy with the work he is doing. Saskatchewan Premier Scott Moe: "We have seen Prime Minister Mark Carney greatly elevate Canada's credibility on the global stage. If we keep looking in the rear view mirror I think we will keep increasingly frustrated and as you look out of the windshield we are seeing some quiet different from this Prime Minister." The lost decade of Justin Trudeau Liberals have created a contrast...but which of these disastrous policies have been removed so far?

Kirk Lubimov

38,850 görüntüleme • 3 ay önce

Today in Ivano-Frankivsk, the final stage of selection for the national Tysiachovesna initiative is taking place. A total of 1,227 projects made it to the final. That is an amazing number. And this number proves that all of our Ukrainian art can become an even greater force. The results of the first Tysiachovesna will now be extremely important: which projects will be created with state support, how they will be received by Ukrainian viewers, listeners, and all our people – the people Tysiachovesna is for. Will Ukrainians in general be interested in this? Will this become a strong voice for Ukraine in the world? This is very important to us, especially in wartime. I believe we will succeed. And exactly one year from now, we will not only be summing up the dry statistics of the first Tysiachovesna through projects that have already been brought to life – we will be talking about the triumphs of our cinema, animation, theater, music, and all of our Ukrainian art, which is here to stay. To be successful, high-quality, and popular among all Ukrainians. This is an important goal for which Tysiachovesna was launched and which it will continue to pursue in the years ahead.

Today in Ivano-Frankivsk, the final stage of selection for the national Tysiachovesna initiative is taking place. A total of 1,227 projects made it to the final. That is an amazing number. And this number proves that all of our Ukrainian art can become an even greater force. The results of the first Tysiachovesna will now be extremely important: which projects will be created with state support, how they will be received by Ukrainian viewers, listeners, and all our people – the people Tysiachovesna is for. Will Ukrainians in general be interested in this? Will this become a strong voice for Ukraine in the world? This is very important to us, especially in wartime. I believe we will succeed. And exactly one year from now, we will not only be summing up the dry statistics of the first Tysiachovesna through projects that have already been brought to life – we will be talking about the triumphs of our cinema, animation, theater, music, and all of our Ukrainian art, which is here to stay. To be successful, high-quality, and popular among all Ukrainians. This is an important goal for which Tysiachovesna was launched and which it will continue to pursue in the years ahead.

Volodymyr Zelenskyy / Володимир Зеленський

286,236 görüntüleme • 19 gün önce

The CBO is wrong: Without this bill, we would face the largest tax hike in history and the CBO would project 1.8% growth. Even with full expensing for manufacturing equipment and historic tax relief, the CBO still projects only 1.8% growth. If you turn up the growth projections to 2.8%, which was achieved during President Trump’s first term, then the debt disappears. We are growing GDP faster than debt and that trend will continue through the remainder of the President’s term.

The CBO is wrong: Without this bill, we would face the largest tax hike in history and the CBO would project 1.8% growth. Even with full expensing for manufacturing equipment and historic tax relief, the CBO still projects only 1.8% growth. If you turn up the growth projections to 2.8%, which was achieved during President Trump’s first term, then the debt disappears. We are growing GDP faster than debt and that trend will continue through the remainder of the President’s term.

Treasury Secretary Scott Bessent

558,485 görüntüleme • 1 yıl önce

In our relentless pursuit of transforming infrastructure, I will like to provide an update on our 18.75km Lekki Epe Expressway Road project. On February 1, 2023, we commissioned phase one, giving the mandate of creating a world-class and long-lasting road to Messrs Craneburg Construction, which starts from Eleko to Epe T-Junction—ensuring a smoother commute for all. Without missing a beat, we embarked on phase two, navigating a challenging stretch from Eleko junction to Abraham Adesanya. This isn't just about roads; it's about reshaping the landscape for a seamless experience. To the resilient residents of Sangotedo, Ogidan, Abijo, Rainoil, and all along Lekki-Epe Expressway, your patience fuels our progress. Please, continue to bear with us as the project is getting to the end. We are promising not just roads but a transformative journey toward enhanced connectivity and convenience. I will continue to provide updates even as we look forward to the completion of this project.

In our relentless pursuit of transforming infrastructure, I will like to provide an update on our 18.75km Lekki Epe Expressway Road project. On February 1, 2023, we commissioned phase one, giving the mandate of creating a world-class and long-lasting road to Messrs Craneburg Construction, which starts from Eleko to Epe T-Junction—ensuring a smoother commute for all. Without missing a beat, we embarked on phase two, navigating a challenging stretch from Eleko junction to Abraham Adesanya. This isn't just about roads; it's about reshaping the landscape for a seamless experience. To the resilient residents of Sangotedo, Ogidan, Abijo, Rainoil, and all along Lekki-Epe Expressway, your patience fuels our progress. Please, continue to bear with us as the project is getting to the end. We are promising not just roads but a transformative journey toward enhanced connectivity and convenience. I will continue to provide updates even as we look forward to the completion of this project.

Babajide Sanwo-Olu

218,176 görüntüleme • 2 yıl önce

Cyprus has now begun its presidency of the Council of the European Union, which will continue through the first half of this year. We are working to make as much progress as possible during this period on opening negotiating clusters and on Ukraine’s accession to the European Union. We discussed this today with the President of Cyprus NikosChristodoulides, and I am counting on productive cooperation. The President and I talked about strengthening sanctions against Russia, which must remain in place as long as Russia’s aggression and occupation continue. We also touched on reinforcement of our air defense, the production and supply of drones, and Cyprus’s participation in the PURL initiative and SAFE instrument. We hope that support for Ukraine will remain strong. Thank you to Cyprus and to the President for the support.

Cyprus has now begun its presidency of the Council of the European Union, which will continue through the first half of this year. We are working to make as much progress as possible during this period on opening negotiating clusters and on Ukraine’s accession to the European Union. We discussed this today with the President of Cyprus NikosChristodoulides, and I am counting on productive cooperation. The President and I talked about strengthening sanctions against Russia, which must remain in place as long as Russia’s aggression and occupation continue. We also touched on reinforcement of our air defense, the production and supply of drones, and Cyprus’s participation in the PURL initiative and SAFE instrument. We hope that support for Ukraine will remain strong. Thank you to Cyprus and to the President for the support.

Volodymyr Zelenskyy / Володимир Зеленський

103,650 görüntüleme • 7 ay önce

And now I’m here, somewhere in England. A shepherd’s hut which will be my home for the next few days while I sit fly fishing on this lake for trout. I don’t know anything about fly fishing, but I’ve bought a rod, reel , flies and even a landing net and I’ve watched a load of YouTube videos, so I’m pretty much an expert.

And now I’m here, somewhere in England. A shepherd’s hut which will be my home for the next few days while I sit fly fishing on this lake for trout. I don’t know anything about fly fishing, but I’ve bought a rod, reel , flies and even a landing net and I’ve watched a load of YouTube videos, so I’m pretty much an expert.

Alastair Hilton

29,016 görüntüleme • 4 ay önce

German chancellor Merz: „I will call the President Trump today. We are preparing on an European level for this meeting (in Alaska) together with the US government so that President Zelenskyy is included. In any case, we cannot accept that in matters of territorial questions talks conducted or even decisions made between the USA and Russia without an European or Ukrainian participation. I assume that the US government shares this view. This is why we have this close discussions. We cannot accept that the entire European security architecture is threatened by Russia. This is why there cannot be peace, if Russia would be rewarded for aggression, which potentially encourages even more aggression. This is our message and I’m grateful that we are in a close dialogue with the Americans. The talks in Alaska next Friday should not be without results. There must be consequences either by increasing pressure on Russia or Russia admits by itself that this war cannot be continued.“

German chancellor Merz: „I will call the President Trump today. We are preparing on an European level for this meeting (in Alaska) together with the US government so that President Zelenskyy is included. In any case, we cannot accept that in matters of territorial questions talks conducted or even decisions made between the USA and Russia without an European or Ukrainian participation. I assume that the US government shares this view. This is why we have this close discussions. We cannot accept that the entire European security architecture is threatened by Russia. This is why there cannot be peace, if Russia would be rewarded for aggression, which potentially encourages even more aggression. This is our message and I’m grateful that we are in a close dialogue with the Americans. The talks in Alaska next Friday should not be without results. There must be consequences either by increasing pressure on Russia or Russia admits by itself that this war cannot be continued.“

(((Tendar)))

379,907 görüntüleme • 1 yıl önce