Qwen3-1.7B

Headlines · 4

1. 4d
   Exowatt builds reusable SageMaker fine-tuning pipeline

   AWS Machine Learning Blog reports that Exowatt used Amazon SageMaker AI serverless model customization to create a repeatable pipeline for task-specific fine-tuned models. A fine-tuned Qwen3-1.7B handled offline alarm triage, while a fine-tuned Qwen3-14B supported cloud energy-market analysis; AWS claims higher accuracy than frontier models at up to 49× lower cost, though those results were not independently verified here. The pipeline turns operational data into a certified training set, evaluated model and deployment artifact, reducing Exowatt's reported workflow from months to days.

   AWS Machine Learning Blog · 🔥 0
2. 6d
   ThinkBit proposes two-bit quantization for reasoning models

   Rhea Sanjay Sukthanker and Pashmina Cameron submitted the ThinkBit paper to arXiv on October 26, 2026. The framework uses layer-specific bit allocation, Gumbel-softmax-based scalar quantization, scale recalibration and quantization-aware training to preserve reasoning while compressing weights to two bits. The authors report that a 0.52 GB Qwen3-1.7B checkpoint scored 46.7 on MATH-500 after recalibration and 73.0 after QAT, while a 3.74 GB model scaled to 14B scored 94.8 versus 97.5 for the 29.5 GB bfloat16 model; these results remain author claims from a single paper.

   Hugging Face · Papers · 🔥 0
3. Oct 17
   RAVEL scores vision encoders for MLLMs without training

   A research team trained and evaluated 210 multimodal large language models (MLLMs) by pairing 70 vision-encoder configurations with Qwen3-1.7B, Qwen2.5-1.5B and SmolLM2-1.7B under a uniform LLaVA-style two-stage recipe, then testing them on 11 downstream benchmarks, according to a report by Machine Heart. They found that as the candidate pool of vision encoders grows more numerous and diverse, existing cross-modal evaluation metrics lose predictive power and can even fall behind purely unimodal vision-only measures. The team's proposed fix, RAVEL, is training-free: it scores a candidate encoder by how consistent image-side and text-side nearest-neighbor relations are, adding stabilized PCA whitening and patch-level similarity matching. The authors report an average Spearman correlation of 0.85 with real downstream performance across the three LLMs, 36.1% above the trained Alignment Probing method and 93.5% and 101.1% above the training-free GW and MutualNN baselines; code, the paper and the 210 MLLM checkpoints are open-sourced.

   机器之心 · 🔥 0
4. Oct 5
   AWS adds SageMaker inference-optimization skill for coding agents

   AWS introduced the aws-ai-ml skill through the Agent Toolkit for AWS, giving MCP-compatible coding agents such as Kiro, Claude Code and Codex SageMaker AI inference optimization and benchmarking expertise. The skill load-tests existing endpoints and reports measured throughput, latency percentiles and concurrency, ranks instance types for models stored in S3, in SageMaker JumpStart or on Hugging Face Hub, compares two benchmark runs, and generates runnable SageMaker Python SDK v3 code. It can be installed locally via an npx command or used in a preconfigured image inside a private Amazon SageMaker Studio JupyterLab space.

   AWS Machine Learning Blog · 🔥 0

Experience and discussion from the community

Share my Qwen3-1.7B experienceAsk about Qwen3-1.7B