Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Alibaba engineer who leads Qwen explained the future of open agent models in 25 minutes - better than $2000 LLM training courses. pre-train the base ->SFT -> RLHF -> tool use -> multi-modal -> ship a whole family (chat / VL / coder / math / QwQ). That loop...

h100envy's profile picture

h100envy

3,743 subscribers

114,423 görüntüleme • 3 ay önce •via X (Twitter)

EğitimBilim & Teknoloji
LIVE

Anya Rossi• Live Now

Private livecam show

13 Yorum

Kirtan profil fotoğrafı
Kirtan3 ay önce

YT link -

Adel Bucetta profil fotoğrafı
Adel Bucetta3 ay önce

that's just product engineering 101, not some magical loop that'll unlock universal ai

sudo rm -rf profil fotoğrafı
sudo rm -rf3 ay önce

And now they don’t release public models.

Kaff 📊 profil fotoğrafı
Kaff 📊3 ay önce

Open agent models are revolutionizing AI

Steven Cheng profil fotoğrafı
Steven Cheng3 ay önce

That pipeline is basically my daily routine with robot prototypes. SFT and RLHF are where the real magic happens, especially for getting consistent tool use in dynamic environments.

TechGeekDavid profil fotoğrafı
TechGeekDavid3 ay önce

Tokenization choices in the base model propagate through every variant in the family. Bad tokenizer puts a ceiling on your coder and math models before fine-tuning starts.

Saeed Anwar profil fotoğrafı
Saeed Anwar3 ay önce

Qwen's real edge is shipping the whole model family at once so devs don't stitch together providers. The pipeline is well understood, execution speed is the differentiator. What's their actual tool use latency compared to GPT function calling in production?

DC｜use.fo profil fotoğrafı
DC｜use.fo3 ay önce

The "pre-train -> SFT -> RLHF" loop is essentially a high-end conveyor belt. Great for scale, but doesn't stop a model from hallucinating with confidence.

Kainoa profil fotoğrafı
Kainoa3 ay önce

I packaged all of this into a Claude Code plugin. One install and your memory, skills, and permissions are configured, no manual setup. Founding price is 19 dollars right now:

AI Mastery Guide profil fotoğrafı
AI Mastery Guide3 ay önce

That whole pipeline explaining Qwen's download numbers makes sense, base to VL to coder to reasoning covers basically every use case people actually have.

Dipanshu Kushwaha profil fotoğrafı
Dipanshu Kushwaha3 ay önce

Nice framework—love the emphasis on the full loop from base to ship. Qwen’s approach is a solid path forward.

John profil fotoğrafı
John3 ay önce

the "ship a whole family" part is the underrated step, everyone skips straight to RLHF and wonders why nothing composes

Piotr Liwszic profil fotoğrafı
Piotr Liwszic3 ay önce

However, this is a very weak model - especially for legal analyses

Benzer Videolar

Prime Intellect engineers explained how they train reasoning models over the open internet in 30 minutes - better than $3000 distributed training courses. split policy and rollouts across nodes -> run agents in parallel envs -> verify with LLM judges -> gradient-sync over the internet -> train Llama, Qwen, Gemma at cluster scale on rented GPUs. That loop is why open reasoning models are catching closed labs without owning a data center. Prime-RL + verifiers + distributed rollouts + LLM judges + multi-cloud GPUs - that's the stack. Watch and save it, then launch your first distributed RL run this week.
30:09

Prime Intellect engineers explained how they train reasoning models over the open internet in 30 minutes - better than $3000 distributed training courses. split policy and rollouts across nodes -> run agents in parallel envs -> verify with LLM judges -> gradient-sync over the internet -> train Llama, Qwen, Gemma at cluster scale on rented GPUs. That loop is why open reasoning models are catching closed labs without owning a data center. Prime-RL + verifiers + distributed rollouts + LLM judges + multi-cloud GPUs - that's the stack. Watch and save it, then launch your first distributed RL run this week.

h100envy

40,022 görüntüleme • 3 ay önce

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.
2:42:11

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

h100envy

497,378 görüntüleme • 3 ay önce

Ex-JPMorgan engineer who wrote the LLM Course explained everything about fine-tuning and merging in 18 minutes - better than $2500 fine-tuning bootcamps. pick the base -> LoRA or QLoRA -> then DPO or KTO for alignment -> merge two fine-tunes into one stronger model -> ship a model that beats the base on your task. That loop is why Labonne's merges sit at the top of the Hugging Face leaderboard. LoRA + QLoRA + DPO + KTO + mergekit - that's the stack. Watch and save it, then merge your first two fine-tunes this week.
17:35

Ex-JPMorgan engineer who wrote the LLM Course explained everything about fine-tuning and merging in 18 minutes - better than $2500 fine-tuning bootcamps. pick the base -> LoRA or QLoRA -> then DPO or KTO for alignment -> merge two fine-tunes into one stronger model -> ship a model that beats the base on your task. That loop is why Labonne's merges sit at the top of the Hugging Face leaderboard. LoRA + QLoRA + DPO + KTO + mergekit - that's the stack. Watch and save it, then merge your first two fine-tunes this week.

h100envy

13,856 görüntüleme • 3 ay önce

Introducing Qwen 3 + Deep Research! We built a deep research agent with Qwen 3 and it performed better than most open source models that we've tried. the code is completely open source and free! built with 's Langgraph Together AI and uses Perplexity Tavily for search here's the code link:
0:57

Introducing Qwen 3 + Deep Research! We built a deep research agent with Qwen 3 and it performed better than most open source models that we've tried. the code is completely open source and free! built with 's Langgraph Together AI and uses Perplexity Tavily for search here's the code link:

Soham

34,604 görüntüleme • 1 yıl önce

... QwQ-Max-Preview Qwen Chat: Blog: 🤔 Today we release "Thinking (QwQ)" in Qwen Chat, backed by our QwQ-Max-Preview, which is a reasoning model based on Qwen2.5-Max. This model is still for preview. It is highly capable of math understanding, coding, agent, etc. Compared with Qwen2.5-Max, it is much smarter and has much more creatitvity. Very soon, we are about to release the official version of QwQ-Max, and we will open-weight both QwQ-Max and Qwen2.5-Max under the license of Apache 2.0! Furthermore, we will also provide smaller variants, e.g., QwQ-32B, which can be deployed on local devices. Also, since a great number of users are expecting our APP, we are going to release an Android and iOS APP while we release our official QwQ-Max. Stay tuned for better things from Qwen!
1:20

... QwQ-Max-Preview Qwen Chat: Blog: 🤔 Today we release "Thinking (QwQ)" in Qwen Chat, backed by our QwQ-Max-Preview, which is a reasoning model based on Qwen2.5-Max. This model is still for preview. It is highly capable of math understanding, coding, agent, etc. Compared with Qwen2.5-Max, it is much smarter and has much more creatitvity. Very soon, we are about to release the official version of QwQ-Max, and we will open-weight both QwQ-Max and Qwen2.5-Max under the license of Apache 2.0! Furthermore, we will also provide smaller variants, e.g., QwQ-32B, which can be deployed on local devices. Also, since a great number of users are expecting our APP, we are going to release an Android and iOS APP while we release our official QwQ-Max. Stay tuned for better things from Qwen!

Qwen

430,876 görüntüleme • 1 yıl önce

Stop scrolling. Qwen Image 2512 is the fastest way to ship clean visuals today. Most people will use it once and miss the real workflow. Do this → → Open Qwen Image 2512 inside Qwen chat → Start with the sample prompts to get baseline results → Generate 3 variations fast (speed is the advantage) → Click Edit and iterate with 1 line instructions → For video, hit Create video and watch the progress bar Save this video, you’ll build better creatives in minutes. Want the SOP? DM me. 💬
38:37

Stop scrolling. Qwen Image 2512 is the fastest way to ship clean visuals today. Most people will use it once and miss the real workflow. Do this → → Open Qwen Image 2512 inside Qwen chat → Start with the sample prompts to get baseline results → Generate 3 variations fast (speed is the advantage) → Click Edit and iterate with 1 line instructions → For video, hit Create video and watch the progress bar Save this video, you’ll build better creatives in minutes. Want the SOP? DM me. 💬

Julian Goldie SEO

10,749 görüntüleme • 9 ay önce

🚀 Exciting News! We're thrilled to announce the launch of Qwen Chat ( ) – your new go-to Web UI for interacting with Qwen models! 🌟 💬 Chat effortlessly with our flagship model Qwen2.5-Plus , explore vision-language capabilities with Qwen2-VL-Max , and dive into reasoning models like QwQ and QVQ, code with coding expert Qwen2.5-Coder-32B-Instruct, etc. ✨ Key Features: * Choose & compare multiple models in one interface * Upload documents & get answers based on them * Preview mode with HTML support * Upload images for visual understanding 🌟 Coming Soon: Web search, image generation, voice mode, and more! 🔗 Ready to experience the future? Start to chat with Qwen today 💬 Feel free to send us your feedback and suggestions!
1:53

🚀 Exciting News! We're thrilled to announce the launch of Qwen Chat ( ) – your new go-to Web UI for interacting with Qwen models! 🌟 💬 Chat effortlessly with our flagship model Qwen2.5-Plus , explore vision-language capabilities with Qwen2-VL-Max , and dive into reasoning models like QwQ and QVQ, code with coding expert Qwen2.5-Coder-32B-Instruct, etc. ✨ Key Features: * Choose & compare multiple models in one interface * Upload documents & get answers based on them * Preview mode with HTML support * Upload images for visual understanding 🌟 Coming Soon: Web search, image generation, voice mode, and more! 🔗 Ready to experience the future? Start to chat with Qwen today 💬 Feel free to send us your feedback and suggestions!

Qwen

215,564 görüntüleme • 1 yıl önce

OpenAI recently released its first open-weights model since GPT-2, entering a field led by DeepSeek and Alibaba's Qwen. Ankit () breaks down these top OSS models, including what sets them apart under the hood: mixture-of-experts, long-context training, and post-training techniques that shape reasoning and alignment—and how different design choices lead to surprisingly similar performance. 00:00 – OpenAI OSS Launch 01:00 – Comparing Open Source LLM Architectures 01:46 – GPT OSS Overview 02:37 – Under The Hood of GPT OSS 03:25 – Qwen-3 Architecture 04:17 – Qwen-3 Training 05:12 – Qwen-3 Post-Training 06:08 – Qwen-3 Reasoning & RL Innovations 06:52 – DeepSeek V3 Overview 07:40 – DeepSeek V3.1 Updates 08:39 – Attention Mechanism (MLA) 09:39 – Comparing Model Sizes 10:35 – Long Context Strategies 11:25 – Reflections on Methods 12:00 – Takeaways
12:31

OpenAI recently released its first open-weights model since GPT-2, entering a field led by DeepSeek and Alibaba's Qwen. Ankit () breaks down these top OSS models, including what sets them apart under the hood: mixture-of-experts, long-context training, and post-training techniques that shape reasoning and alignment—and how different design choices lead to surprisingly similar performance. 00:00 – OpenAI OSS Launch 01:00 – Comparing Open Source LLM Architectures 01:46 – GPT OSS Overview 02:37 – Under The Hood of GPT OSS 03:25 – Qwen-3 Architecture 04:17 – Qwen-3 Training 05:12 – Qwen-3 Post-Training 06:08 – Qwen-3 Reasoning & RL Innovations 06:52 – DeepSeek V3 Overview 07:40 – DeepSeek V3.1 Updates 08:39 – Attention Mechanism (MLA) 09:39 – Comparing Model Sizes 10:35 – Long Context Strategies 11:25 – Reflections on Methods 12:00 – Takeaways

Y Combinator

208,828 görüntüleme • 1 yıl önce

Liquid AI's head of post-training explained how they built a small model that runs on-device under 1 GB in 20 minutes - better than $2500 small-model bootcamps. pick LFM2.5 base -> on-policy preference alignment -> agentic reinforcement learning -> curriculum training -> iterative model merging -> ship a 1B model that reliably calls tools on your phone. That loop is why frontier small models now beat 70B models on the tasks that actually matter. LFM2.5 + on-policy DPO + agentic RL + curriculum training + iterative merging - that's the stack. Watch and save it, then run a 1B agent on your phone tonight.
19:11

Liquid AI's head of post-training explained how they built a small model that runs on-device under 1 GB in 20 minutes - better than $2500 small-model bootcamps. pick LFM2.5 base -> on-policy preference alignment -> agentic reinforcement learning -> curriculum training -> iterative model merging -> ship a 1B model that reliably calls tools on your phone. That loop is why frontier small models now beat 70B models on the tasks that actually matter. LFM2.5 + on-policy DPO + agentic RL + curriculum training + iterative merging - that's the stack. Watch and save it, then run a 1B agent on your phone tonight.

h100envy

87,975 görüntüleme • 2 ay önce

Qwopus 3.6-27b Coder I had a lot of requests to test it, so I did. I ran the same tests I’ve done on other models. It scored lower on tool-eval-bench than both Qwen 3.6 27b and Qwen 3.6 35b. It also produced broken single-file Tetris and Solar System HTML demos, even after 3 fresh attempts. In the Tetris html, the game stalled when you completed a line. It also randomly changed the pieces when moving them left or right. In the Solar System html, clicking on the sun or any planet broke the rendering — the objects just disappeared. Conclusion: The regular Qwen 3.6 27b and 35b models performed better. What do you think of Qwopus 3.6-27b Coder? Full results & RAW files: Full prompts in the posts below.
0:46

Qwopus 3.6-27b Coder I had a lot of requests to test it, so I did. I ran the same tests I’ve done on other models. It scored lower on tool-eval-bench than both Qwen 3.6 27b and Qwen 3.6 35b. It also produced broken single-file Tetris and Solar System HTML demos, even after 3 fresh attempts. In the Tetris html, the game stalled when you completed a line. It also randomly changed the pieces when moving them left or right. In the Solar System html, clicking on the sun or any planet broke the rendering — the objects just disappeared. Conclusion: The regular Qwen 3.6 27b and 35b models performed better. What do you think of Qwopus 3.6-27b Coder? Full results & RAW files: Full prompts in the posts below.

Mia

11,183 görüntüleme • 3 ay önce

Ex-Google engineer explained AI agent loops, harness, evals in 20 minutes - better than 500$ courses. trace every run → judge it with an LLM → diagnose → fix → ship. That loop is how agents self-improve over time. Agent loops + memory + harness + evals - thats the stack. Watch it, then save the framework below.
19:30

Ex-Google engineer explained AI agent loops, harness, evals in 20 minutes - better than 500$ courses. trace every run → judge it with an LLM → diagnose → fix → ship. That loop is how agents self-improve over time. Agent loops + memory + harness + evals - thats the stack. Watch it, then save the framework below.

Movez

567,575 görüntüleme • 3 ay önce

Introducing Jan-v2-VL, a multimodal agent built for long-horizon tasks. Jan-v2-VL executes 49 steps without failure, while the base model stops at 5 and other similar-scale VLMs stop between 1 and 2. It achieves longer, stable task execution in your browser without accuracy loss. 3 variants are available: - Jan-v2-VL-low (efficiency-oriented) - Jan-v2-VL-med (balanced) - Jan-v2-VL-high (deeper reasoning and longer execution) Models: To use it, update your Jan App and download Jan-v2-VL from the Model Hub. Activate Browser MCP servers for agentic use cases. Credit to the Qwen team for Qwen3-VL-8B-Thinking base model.
0:52

Introducing Jan-v2-VL, a multimodal agent built for long-horizon tasks. Jan-v2-VL executes 49 steps without failure, while the base model stops at 5 and other similar-scale VLMs stop between 1 and 2. It achieves longer, stable task execution in your browser without accuracy loss. 3 variants are available: - Jan-v2-VL-low (efficiency-oriented) - Jan-v2-VL-med (balanced) - Jan-v2-VL-high (deeper reasoning and longer execution) Models: To use it, update your Jan App and download Jan-v2-VL from the Model Hub. Activate Browser MCP servers for agentic use cases. Credit to the Qwen team for Qwen3-VL-8B-Thinking base model.

👋 Jan

130,375 görüntüleme • 11 ay önce

Ex-Google engineer reveals how to build AI agent loops, harnesses, LLM ops, and evals in 19 minutes. Trace → evaluate → diagnose → fix → ship → repeat. That loop is how agents self-improve over time. Agentic loops + harness + memory + evals - that’s the senior engineer stack. This is better than $500 paid courses on the same topic, explained in under 20 minutes. Watch it, then save the framework below.
20:10

Ex-Google engineer reveals how to build AI agent loops, harnesses, LLM ops, and evals in 19 minutes. Trace → evaluate → diagnose → fix → ship → repeat. That loop is how agents self-improve over time. Agentic loops + harness + memory + evals - that’s the senior engineer stack. This is better than $500 paid courses on the same topic, explained in under 20 minutes. Watch it, then save the framework below.

Codez

173,520 görüntüleme • 3 ay önce

Ex-Berkeley PhD who leads SGLang at xAI explained how they serve Grok on 100K GPUs in 23 minutes - better than $2000 inference-at-scale courses. split prefill and decode -> shard experts across GPUs -> route tokens per expert -> overlap comm and compute -> serve at DeepSeek-API-killing prices. That loop is why xAI runs Grok on SGLang and third parties beat DeepSeek's own API by 5x on cost. SGLang + prefill-decode disaggregation + expert parallelism + AMD MI300 - that's the stack. Watch and save it, then read the article below.
22:47

Ex-Berkeley PhD who leads SGLang at xAI explained how they serve Grok on 100K GPUs in 23 minutes - better than $2000 inference-at-scale courses. split prefill and decode -> shard experts across GPUs -> route tokens per expert -> overlap comm and compute -> serve at DeepSeek-API-killing prices. That loop is why xAI runs Grok on SGLang and third parties beat DeepSeek's own API by 5x on cost. SGLang + prefill-decode disaggregation + expert parallelism + AMD MI300 - that's the stack. Watch and save it, then read the article below.

h100envy

319,581 görüntüleme • 3 ay önce

OpenAI engineer explained every form of fine-tuning in 1 hour 46 minutes - better than $3000 fine-tuning bootcamps. pick your task -> try prompting first -> drop to SFT for style -> run DPO for preference -> run RFT for verifiable rewards -> ship a model 10x cheaper than GPT-5. That loop is why teams that fine-tune beat teams that only prompt on cost and speed. Prompting + SFT + DPO + RFT + OpenAI fine-tuning API - that's the stack. Watch and save it, then pick the right fine-tune for your task this week.
1:45:58

OpenAI engineer explained every form of fine-tuning in 1 hour 46 minutes - better than $3000 fine-tuning bootcamps. pick your task -> try prompting first -> drop to SFT for style -> run DPO for preference -> run RFT for verifiable rewards -> ship a model 10x cheaper than GPT-5. That loop is why teams that fine-tune beat teams that only prompt on cost and speed. Prompting + SFT + DPO + RFT + OpenAI fine-tuning API - that's the stack. Watch and save it, then pick the right fine-tune for your task this week.

h100envy

41,245 görüntüleme • 2 ay önce

Modal engineer who teaches Full-Stack Deep Learning explained everything an AI engineer needs to know about GPUs in 20 minutes - better than $800 CUDA workshops. memory bandwidth -> SM occupancy -> kernel fusion -> arithmetic intensity -> know when you're compute-bound vs memory-bound. That loop is how you stop overpaying for GPUs you're barely using. VRAM math + roofline model + Modal serverless GPUs + kernel profiling - that's the stack. Watch and save it, then check your own GPU utilization tonight.
19:25

Modal engineer who teaches Full-Stack Deep Learning explained everything an AI engineer needs to know about GPUs in 20 minutes - better than $800 CUDA workshops. memory bandwidth -> SM occupancy -> kernel fusion -> arithmetic intensity -> know when you're compute-bound vs memory-bound. That loop is how you stop overpaying for GPUs you're barely using. VRAM math + roofline model + Modal serverless GPUs + kernel profiling - that's the stack. Watch and save it, then check your own GPU utilization tonight.

h100envy

18,442 görüntüleme • 3 ay önce

Qwopus on a single RTX 3090. Claude Opus 4.6 reasoning distilled into Qwen 3.5 27B dense, running through Claude's own coding agent (claude code). 29-35 tok/s with thinking mode on. the jinja bug that kills thinking on base Qwen doesn't carry over. harness and model matched. the base model would pause mid task on Claude Code. just stop generating. that's why i ran it through OpenCode, which handles stalled states automatically. this distilled version doesn't stall. it waits for tool outputs, reads them, selfcorrects when something breaks, and keeps going. i gave it a benchmark analysis task. went 9 minutes autonomous. wrote a README nobody asked for. zero steering. video is 5x speed but fully uncut. if you have a 3090, you can run this right now. free. no API. no subscription. opus structured reasoning on localhost. octopus invaders is next. same prompt that base qwen passed in 13 minutes and hermes 4.3 failed on 2x the hardware. i want to see if the distillation changes the outcome or just the style. more data soon.
1:53

Qwopus on a single RTX 3090. Claude Opus 4.6 reasoning distilled into Qwen 3.5 27B dense, running through Claude's own coding agent (claude code). 29-35 tok/s with thinking mode on. the jinja bug that kills thinking on base Qwen doesn't carry over. harness and model matched. the base model would pause mid task on Claude Code. just stop generating. that's why i ran it through OpenCode, which handles stalled states automatically. this distilled version doesn't stall. it waits for tool outputs, reads them, selfcorrects when something breaks, and keeps going. i gave it a benchmark analysis task. went 9 minutes autonomous. wrote a README nobody asked for. zero steering. video is 5x speed but fully uncut. if you have a 3090, you can run this right now. free. no API. no subscription. opus structured reasoning on localhost. octopus invaders is next. same prompt that base qwen passed in 13 minutes and hermes 4.3 failed on 2x the hardware. i want to see if the distillation changes the outcome or just the style. more data soon.

Sudo su

296,392 görüntüleme • 7 ay önce

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.
0:44

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.

stevibe

22,882 görüntüleme • 5 ay önce