HF Jobs runs vLLM per-minute, Ai2 ships Olmo diagnostic, Datasette plugin unpins
Every URL the pipeline pulled into ranking for this issue — primary sources plus the supporting and contradicting findings each Researcher returned. Inline citations in the issue point back here.
Sources
Which tokens does a hybrid model predict better? huggingface.co
Run a vLLM Server on HF Jobs in One Command huggingface.co
datasette-export-database 0.3a2 simonwillison.net
Release: datasette-export-database 0.3a2 An embarrassingly tiny release. The pyproject.toml had pinned to datasette==1.0a27 , inadvertently making this plugin incompatible with all other Datasette versions. It’s now datasette>=1.0a27 instead. Tags: datasette
References
Michael Hannecke, Medium (vLLM Security Hardening Guide) medium.com
vLLM’s —api-key flag only secures specific /v1 endpoints, leaving other critical diagnostic or monitoring paths completely unauthenticated… CVE-2026-22778 (CVSS 9.8) allowed for Remote Code Execution by sending malicious URLs to video-processing endpoints that lacked authentication.
JarvisLabs blog — Scaling LLM Inference (DP/PP/TP) jarvislabs.ai
Increasing to TP=8 on a single node can actually lead to a throughput drop under high concurrency compared to TP=4… frequent NCCL all-reduce calls saturate the interconnect bandwidth, especially on PCIe-based systems without NVLink.
JarvisLabs — H200 price guide jarvislabs.ai
Specialized GPU clouds offer on-demand H200 instances ranging from $1.40 to $3.80 per hour… self-hosting on a specialized cloud can save between $3,000 and $5,000 per month compared to managed services like HF Dedicated Endpoints.
Hugging Face blog — HF CLI for Agents huggingface.co
The CLI is now ‘agent-optimized,’ detecting when it is being driven by AI tools like Claude Code or Codex… structured hints that suggest the next logical command with pre-filled IDs, which reportedly reduces token usage by up to 6x compared to manual API calls.
Spheron Network — HF Inference Endpoints alternatives spheron.network
Waking an endpoint can take several minutes as weights are pulled and the container initializes… this delay makes scale-to-zero unsuitable for real-time, user-facing applications unless paired with a custom client-side queuing system.
Cyfuture Cloud — H200 price/specs guide cyfuture.cloud
The H200’s 141GB HBM3e memory is nearly double the H100’s 80GB capacity… H200 can deliver up to 2x faster inference speeds for memory-bound tasks, potentially lowering the total cost-per-token despite a higher hourly rental price.
Jelassi et al., ‘Repeat After Me’ (arXiv 2406.07522) arxiv.org
Vanilla Mamba models fail to reliably copy tokens once sequences exceed roughly 2,000 tokens, whereas Transformers maintain near-perfect accuracy… for Mamba to perform a perfect COPY operation on arbitrary sequences, its state size would need to grow linearly with the input length.
Finite-precision LRNNs require negative eigenvalues in the range [-1, 1] to flip bits or track state changes effectively… incorporating negative values allows these architectures to simulate any finite state automaton.
Microsoft Research, ‘Samba: Simple Hybrid State Space Models’ microsoft.com
Samba pairs Mamba with Sliding Window Attention and has demonstrated ‘perfect memory recall’ on the Passkey Retrieval and Phonebook tasks… tested up to 1M tokens in zero-shot settings.
Hugging Face blog (bird-of-paradise), ‘Three Horsemen of Numerical Divergence’ huggingface.co
Because recurrent states are highly sensitive to computational precision, practitioners must often use FP32 precision for hidden states and disable certain optimizations like ‘cascade attention’ to avoid reward collapse during RL.
Medium, ‘Why Scaling Pre-training Loss Might Be Ruining Your LLMs Reasoning’ medium.com
Two models achieving the same validation loss can exhibit vastly different capabilities… a model might achieve a low average loss by being highly accurate on ‘easy’ frequent tokens while remaining incapable of the complex ‘long-tail’ reasoning required for specialized tasks.
DeepLearning.AI, ‘AI21 Labs Jamba 1.5 Outpaces Transformers in Long-Text Processing’ deeplearning.ai
Jamba supports a 256K context window on a single GPU and achieves 3x the throughput of models like Mixtral-8x7B… interleaving attention and Mamba layers at a 1:7 ratio to balance reasoning and throughput.
letsdatascience.com (Datasette 1.0 alpha status) letsdatascience.com
As of late June 2026, the project has reached version 1.0a35, introducing major features like a native web interface for creating and altering tables… installing plugins via standard commands sometimes fails because many are pinned to specific alpha versions to accommodate evolving hook signatures.
