https://pinboard.in/u:jm/public/ recent bookmarks from jm 2025-07-18T08:52:31+00:00 https://prospect.org/power/2025-07-17-how-did-elon-musk-turn-grok-into-mechahitler/ jmLLMs become “woke” because they are trained to be pro-social — to be helpful, kindly, truthful, and not to say bigoted or cruel things. Training it to do the opposite — to be anti-woke — is to activate every antisocial association in the English language, including racism, sexism, cruelty, dishonesty, and Nazism. According to a vast statistical representation of the English language constructed by none other than Elon Musk, that’s what anti-wokeness is. “Elon Musk is repeatedly insisting, no, no, there’s a difference between what I’m doing and being a Nazi. And what the model keeps telling him is, statistically, that’s not the case,” said Schou. A key implication here is that LLMs will tend to converge on similar types of behavior. The above researchers were not using Grok, but they found the exact same pattern of powerful association groupings of good and evil in other LLMs — and these can’t be removed through fine-tuning. One could imagine the RLHF process including adjustments of every parameter, but experts said that this will degrade or break the model. The matrices in an LLM are arranged hierarchically, and the top layers get fixed in place relatively early in pretraining. Mess with them, and the model will stop working. Instead, RLHF developed more like a series of gates that prevent undesired outcomes. “The model completes most of the computation that it needs in order to reach a particular outcome,” [Andreas] Schou said. “And then says, ‘Wait, wait, wait, I’m saying that I’m MechaHitler. No, I’m not doing that.’” One could try to assemble a custom dataset with nothing but “conservatism minus the Nazis” and train a new model from scratch, but not only would that be extremely expensive, it also would not be nearly as strong as leading models, since its universe of available training data would be much smaller. Funnily enough, the latter approach is exactly what Elon Musk claims xAI are now doing.]]> llms language technology ai mechahitler grok tuning rlhf woke training https://pinboard.in/ https://pinboard.in/u:jm/b:08edfa8b6a64/ 2023-01-05T15:53:26+00:00 https://krzysztofslusarski.github.io/2022/12/12/async-manual.html jm async-profiler java jvm profiling tracing performance tuning https://pinboard.in/ https://pinboard.in/u:jm/b:8ec959433ae2/ 2021-05-21T08:54:01+00:00 https://talawah.io/blog/extreme-http-performance-tuning-one-point-two-million/ jm http servers c10k linux performance scalability ops tuning libreactor networking tcp https://pinboard.in/ https://pinboard.in/u:jm/b:022774113a4b/ 2020-12-04T10:57:55+00:00 https://gceasy.io/ jm java gc tuning performance jvm logs ops https://pinboard.in/ https://pinboard.in/u:jm/b:d3d8327d2376/ 2020-03-11T13:21:50+00:00 https://medium.com/thron-tech/lessons-learned-about-monitoring-the-jvm-in-the-era-of-containers-47e7fe0b77dc jm jvm g1 g1gc gc tuning metrics ops https://pinboard.in/ https://pinboard.in/u:jm/b:44d5b27b29bf/ 2019-04-10T09:48:34+00:00 https://aws.amazon.com/blogs/big-data/best-practices-for-successfully-managing-memory-for-apache-spark-applications-on-amazon-emr/ jm spark emr aws tuning memory ooms java https://pinboard.in/ https://pinboard.in/u:jm/b:7d538a70d00d/ 2019-03-25T12:37:07+00:00 https://devcenter.heroku.com/articles/tuning-glibc-memory-behavior#what-value-to-choose-for-malloc_arena_max jm ops ruby memory malloc allocation arenas tuning fragmentation https://pinboard.in/ https://pinboard.in/u:jm/b:f6214c1a141c/ 2016-06-29T16:30:13+00:00 http://www.brendangregg.com/USEmethod/use-linux.html jm linux sar iostat mpstat cli ops sysadmin performance tuning use metrics https://pinboard.in/ https://pinboard.in/u:jm/b:c50f8a259903/ 2016-06-18T09:46:44+00:00 https://blog.buoyant.io/2016/06/17/small-memory-jvm-techniques-for-microservice-sidecars/ jmWe found two specific techniques to be the most beneficial: turning off one of the two JIT compilers enabled by default (the “C2” compiler), and using a 32-bit, rather than a 64-bit, JVM. ]]> 32bit jvm java ops memory tuning jit linkerd https://pinboard.in/ https://pinboard.in/u:jm/b:ade3b2e45f61/ 2016-04-13T16:00:35+00:00 https://engineering.linkedin.com/blog/2016/04/dr-elephant-open-source-self-serve-performance-tuning-hadoop-spark jm[LinkedIn] are proud to announce today that we are open sourcing Dr. Elephant, a powerful tool that helps users of Hadoop and Spark understand, analyze, and improve the performance of their flows. neat, although I've been bitten too many times by LinkedIn OSS release quality at this point to jump in.... ]]> linkedin oss hadoop spark performance tuning ops https://pinboard.in/ https://pinboard.in/u:jm/b:48eaac9e2c7d/ 2016-02-13T22:01:51+00:00 https://twitter.com/akpurtell/status/698614296226312192 jm optimization funny terms terminology graduate-student-descent hyperparameters semisupervised-intern-descent tuning interns https://pinboard.in/ https://pinboard.in/u:jm/b:5eb66f4c558b/ 2015-09-28T15:30:07+00:00 http://blog.threatstack.com/scaling-cassandra-lessons-learned jm cassandra configuration tuning scale ops https://pinboard.in/ https://pinboard.in/u:jm/b:41bfcab2a82b/ 2015-06-06T07:02:50+00:00 https://databricks.com/blog/2015/05/28/tuning-java-garbage-collection-for-spark-applications.html jm g1gc gc java jvm performance spark ops tuning https://pinboard.in/ https://pinboard.in/u:jm/b:431050bae8c9/ 2015-05-09T08:19:41+00:00 http://www.slideshare.net/MonicaBeckwith/garbage-first-garbage-collector-g1-gc-migration-to-expectations-and-advanced-tuning jm gc g1gc jvm java tuning performance ops migration https://pinboard.in/ https://pinboard.in/u:jm/b:1dbd33f78572/ 2015-04-29T15:42:34+00:00 https://issues.apache.org/jira/browse/CASSANDRA-7486 jm cassandra tuning ops g1gc cms gc java jvm production performance memory https://pinboard.in/ https://pinboard.in/u:jm/b:0b4b3d1abe38/ 2015-04-11T20:21:41+00:00 http://engineering.linkedin.com/java/optimizing-java-cms-garbage-collections-its-difficulties-and-using-jtune-solution jm java jvm tuning gc cms linkedin performance ops https://pinboard.in/ https://pinboard.in/u:jm/b:717dfe95f072/ 2015-04-06T20:27:37+00:00 https://groups.google.com/d/msg/mechanical-sympathy/DWlziVmyW-w/at-54WECjL4J jmBased on empirical evidence (across many tens of sites thus far) and note-comparing with others, I use a list of "usual suspects" that I blame whenever they are not set to my liking and system-level hiccups are detected. Getting these settings right from the start often saves a bunch of playing around (and no, there is no "priority" to this - you should set them all right before looking for more advice...). ]]> performance latency hiccups gil-tene tuning mechanical-sympathy hyperthreading linux ops https://pinboard.in/ https://pinboard.in/u:jm/b:e22d85fc3109/ 2015-02-26T15:49:43+00:00 http://www.jclarity.com/2015/02/24/why-we-built-illuminate-where-apm-is-going-next/ jmUsers download and install an Illuminate Daemon using a simple installer which starts up a small stand alone Java process. The Daemon sits quietly unless it is asked to start gathering SLA data and/or to trigger a diagnosis. Users can set SLA’s via the dashboard and can opt to collect latency measurements of their transactions manually (using our library) or by asking Illuminate to automatically instrument their code (Servlet and JDBC based transactions are currently supported). SLA latency data for transactions is collected on a short cycle. When the moving average of latency measurements goes above the SLA value (e.g. 150ms), a diagnosis is triggered. The diagnosis is very quick, gathering key data from O/S, JVM(s), virtualisation and other areas of the system. The data is then run through the machine learned algorithm which will quickly narrow down the possible causes and gather a little extra data if needed. Once Illuminate has determined the root cause of the performance problem, the diagnosis report is sent back to the dashboard and an alert is sent to the user. That alert contains a link to the result of the diagnosis which the user can share with colleagues. Illuminate has all sorts of backoff strategies to ensure that users don’t get too many alerts of the same type in rapid succession! ]]> illuminate jclarity java jvm scala latency gc tuning performance https://pinboard.in/ https://pinboard.in/u:jm/b:d8f4dcc75538/ 2015-02-03T22:09:12+00:00 http://tobert.github.io/tldr/cassandra-java-huge-pages.html jm jvm performance tuning huge-pages vm ops cassandra java https://pinboard.in/ https://pinboard.in/u:jm/b:d448b5d53e34/ 2015-01-25T23:43:59+00:00 https://speakerdeck.com/elasticsearch/maintaining-performance-in-distributed-systems jm performance elasticsearch java jvm ops tuning https://pinboard.in/ https://pinboard.in/u:jm/b:99d8451bc962/ 2014-07-11T13:54:10+00:00 https://twitter.com/AWSSummits/status/487280813780656129/photo/1 jm Cheat sheet of EBS-optimized instances. http://t.co/vmTlhUtpWk Optimize your queue depth to achieve lower latency & highest IOPS. http://t.co/EO48oa0D6X When configuring your RAID, use a stripe size of 128KB or 256KB. http://t.co/N0ldtFJ4t6 Use larger block size to speed up the pre-warming process. http://t.co/8UoIeWE2px ]]> ebs aws amazon iops raid ops tuning https://pinboard.in/ https://pinboard.in/u:jm/b:dbfd882c9ad6/ 2014-06-17T11:19:59+00:00 http://aws.amazon.com/articles/3604 jm aws sdk java logging ec2 s3 dynamodb sockets tuning https://pinboard.in/ https://pinboard.in/u:jm/b:6e67b33f2691/ 2014-05-29T13:53:01+00:00 http://docs.basho.com/riak/latest/ops/advanced/backends/leveldb/ jm leveldb tuning performance ops riak https://pinboard.in/ https://pinboard.in/u:jm/b:8dffa2129e4d/ 2014-05-01T22:49:27+00:00 http://vincent.bernat.im/en/blog/2014-tcp-time-wait-state-linux.html jm networking linux tcp performance time-wait sysctls tuning https://pinboard.in/ https://pinboard.in/u:jm/b:1161c9971396/ 2014-04-30T10:57:02+00:00 https://speakerdeck.com/pyconslides/scaling-realtime-at-disqus-by-adam-hitchcock jm sysctl nginx tuning go disqus realtime push eventsource websockets long-polling iframe python https://pinboard.in/ https://pinboard.in/u:jm/b:92ccc8b3f3ef/ 2014-04-08T21:54:13+00:00 http://engineering.linkedin.com/garbage-collection/garbage-collection-optimization-high-throughput-and-low-latency-java-applications jm performance optimization linkedin java jvm gc tuning https://pinboard.in/ https://pinboard.in/u:jm/b:77cabe371f7c/ 2014-03-21T11:26:50+00:00 http://vanillajava.blogspot.co.uk/2013/07/micro-jitter-busy-waiting-and-binding.html?m=1 jm pinning threads performance latency jitter tuning https://pinboard.in/ https://pinboard.in/u:jm/b:186b39854e95/ 2014-03-19T12:12:51+00:00 http://java-is-the-new-c.blogspot.ie/2013/07/impact-of-large-primitive-arrays-blobs.html jm cms java jvm performance optimization tuning off-heap-storage memory https://pinboard.in/ https://pinboard.in/u:jm/b:026572f118c8/ 2014-02-12T10:49:07+00:00 http://www.fullcontact.com/blog/mongo-to-cassandra-migration/ jm cassandra luks raid linux tuning ops blockdev disks sdd https://pinboard.in/ https://pinboard.in/u:jm/b:9e9bd44bf027/ 2014-01-24T10:14:24+00:00 http://tech.shift.com/post/74311817513/cassandra-tuning-the-jvm-for-read-heavy-workloads jmThe cluster we tuned is hosted on AWS and is comprised of 6 hi1.4xlarge EC2 instances, with 2 1TB SSDs raided together in a raid 0 configuration. The cluster’s dataset is growing steadily. At the time of this writing, our dataset is 341GB, up from less than 200GB a few months ago, and is growing by 2-3GB per day. The workload on this cluster is very read heavy, with quorum reads making up 99% of all operations. Some careful GC tuning here. Probably not applicable to anyone else, but good approach in general.]]> java performance jvm scaling gc tuning cassandra ops https://pinboard.in/ https://pinboard.in/u:jm/b:bc4cff6803b7/ 2014-01-03T14:19:46+00:00 http://engineering.chartbeat.com/2014/01/02/part-1-lessons-learned-tuning-tcp-and-nginx-in-ec2/ jm nginx amazon ec2 tcp ip tuning sysctl linux c10k ssl http https://pinboard.in/ https://pinboard.in/u:jm/b:4c04f47e8135/ 2013-12-21T20:50:48+00:00 http://www.slideshare.net/cpwatson/cpn302-yourlinuxamioptimizationandperformance jm netflix aws tuning ami perf systemtap tunables sched_batch batch hadoop optimization performance https://pinboard.in/ https://pinboard.in/u:jm/b:5c1292b5cc15/ 2013-11-10T00:10:17+00:00 http://blog.ragozin.info/2013/11/hotspot-jvm-garbage-collection-options.html jm jvm hotspot java tuning ops gc https://pinboard.in/ https://pinboard.in/u:jm/b:a6b132013693/ 2013-10-23T16:28:35+00:00 http://kerneldave.blogspot.ie/2013/10/response-to-optimizing-linux-memory.html jmDo not read in to this article too much, especially for trying to understand how the Linux VM or the kernel works.  The authors misread the "global spinlock on the zone" source code and the interpretation in the article is dead wrong. ]]> linux tuning vm kernel linkedin memory numa https://pinboard.in/ https://pinboard.in/u:jm/b:028705973c7a/ 2013-10-08T11:15:46+00:00 http://java-is-the-new-c.blogspot.de/2013/07/what-drives-full-gc-duration-its.html jmFull GC duration depends on the number of objects allocated and the locality of their references. It does not depend that much on actual heap size. Reference locality has a surprisingly high effect.]]> java jvm data gc tuning performance cms g1 https://pinboard.in/ https://pinboard.in/u:jm/b:71efc8abad7c/ 2013-09-18T21:34:06+00:00 http://www.infoq.com/articles/tuning-tips-G1-GC jmg1gc gc java jvm tuning ops optimization https://pinboard.in/ https://pinboard.in/u:jm/b:6b12d8d218aa/ 2013-09-04T14:26:32+00:00 http://www.slideshare.net/amywtang/wbdb2012-voldemortssd jm With SSD, we find that garbage collection will become a very significant bottleneck, especially for systems which have little control over the storage layer and rely on Java memory management. Big heapsizes make the cost of garbage collection expensive, especially the single threaded CMS Initial mark. We believe that data systems must revisit their caching strategies with SSDs. In this regard, SSD has provided an efficient solution for handling fragmentation and moving towards predictable multitenancy.]]> voldemort storage ssd disk linkedin big-data jvm tuning ops gc https://pinboard.in/ https://pinboard.in/u:jm/b:086f50102a3b/ 2013-07-25T11:43:37+00:00 http://www.jclarity.com/products/censum/ jm[JVM] GC is a difficult, specialised area that can be very frustrating for busy developers or devops folks to deal with. The JVM has a number of Garbage Collectors and a bewildering array of switches that can alter the behaviour of each collector. Censum does all of the parsing, number crunching and statistical analysis for you, so you don't have to go and get that PhD in Computer Science in order to solve your GC performance problem. Censum gives you straight answers as opposed to a ton of raw data. can eat any GC log you care to throw at it. is easy to install and use. Commercial software, UKP 495 per license.]]> censum gc tuning ops java jvm commercial https://pinboard.in/ https://pinboard.in/u:jm/b:ed538dc1a0a6/ 2013-07-25T11:27:30+00:00 http://java-is-the-new-c.blogspot.ie/2013/07/tuning-and-benchmarking-java-7s-garbage.html jm java gc tuning jvm cms g1 ops https://pinboard.in/ https://pinboard.in/u:jm/b:6ef833ae0fa3/ 2013-07-16T22:22:21+00:00 http://mechanical-sympathy.blogspot.ca/2013/07/java-garbage-collection-distilled.html jm jvm java gc garbage-collection tuning memory performance martin-thompson https://pinboard.in/ https://pinboard.in/u:jm/b:f32e30988004/ 2013-05-14T20:42:42+00:00 http://distributeddreams.blogspot.ie/2012/04/making-sense-out-of-bdb-je-fast-stats.html jm voldemort caching bdb bdb-je storage tuning ops metrics reference https://pinboard.in/ https://pinboard.in/u:jm/b:14b3ea9af8a3/ 2013-04-05T16:15:57+00:00 http://www.psc.edu/index.php/networking/641-tcp-tune jmThese notes are intended to help users and system administrators maximize TCP/IP performance on their computer systems. They summarize all of the end-system (computer system) network tuning issues including a tutorial on TCP tuning, easy configuration checks for non-experts, and a repository of operating system specific instructions for getting the best possible network performance on these platforms. Some tips for maximizing HPC network performance for the intra-DC case; recommended by the LinkedIn Kafka operations page.]]> tuning network tcp sysadmin performance ops kafka ec2 https://pinboard.in/ https://pinboard.in/u:jm/b:90ef24f751cb/ 2013-02-28T21:45:40+00:00 http://imho.nu/iMHo/tcp-optimizer/ jmTCP optimizer modifies a few settings inside iOS, including increasing the TCP receive buffer from 131072 to 292000, disabling TCP delayed ACK’s, allowing a maximum of 16 un-ACK’d packets instead of 8 and set the default packet size to 1460 instead of 512. These changes won’t only speed up your YouTube videos, they’ll also improve your internet connection’s performance overall, including Wi-Fi network connectivity. ]]> tcp performance tuning ios apple wifi wireless 802.11n sysctl ip https://pinboard.in/ https://pinboard.in/u:jm/b:96226cdb6566/ 2012-12-01T18:51:22+00:00 http://www.slideshare.net/aszegedi/everything-i-ever-learned-about-jvm-performance-tuning-twitter jm tuning jvm java gc cms presentations slides twitter https://pinboard.in/ https://pinboard.in/u:jm/b:80c49db84474/ 2012-07-23T10:05:43+00:00 http://blog.explainmydata.com/2012/07/expensive-lessons-in-python-performance.html jm picloud prun guppy namedtuples python optimization performance tuning profiling https://pinboard.in/ https://pinboard.in/u:jm/b:78e224a43e8c/ 2011-12-04T23:00:35+00:00 http://www.slideshare.net/boundary_slides/garbage-garbage-everywhere jm gc java jvm event-streams event-processing tuning slides presentations scott-andreas performance https://pinboard.in/ https://pinboard.in/u:jm/b:750e291a35af/ 2010-09-29T15:48:24+00:00 http://jcole.us/blog/archives/2010/09/28/mysql-swap-insanity-and-the-numa-architecture/ jm linux memory mysql optimization performance swap tuning vm numa swap-insanity swapping https://pinboard.in/u:jm/b:6aad45b4a7bc/ 2010-02-19T11:55:55+00:00 http://blip.tv/file/2232410 jm python gil locking synchronization ouch performance tuning coding interpreters threads pthreads https://pinboard.in/u:jm/b:005314349ca9/