@HanchungLee Avatar @HanchungLee Han

Han posts on X about open ai, ai, $googl, microsoft the most. They currently have [-----] followers and [----] posts still getting attention that total [------] engagements in the last [--] hours.

Engagements: [------] #

Engagements Line Chart

Mentions: [--] #

Mentions Line Chart

Followers: [-----] #

Followers Line Chart

CreatorRank: [-------] #

CreatorRank Line Chart

Social Influence

Social category influence technology brands stocks finance social networks countries travel destinations celebrities fashion brands automotive brands currencies

Social topic influence open ai, ai, $googl, microsoft, google, china, databricks, llm, meta, facebook

Top assets mentioned Alphabet Inc Class A (GOOGL) Microsoft Corp. (MSFT) GrokCoin (GROKCOIN) DeepSeek (DEEPSEEK) IBM (IBM) Tesla, Inc. (TSLA) NVIDIA Corp. (NVDA) Broadcom, Inc. (AVGO) Costco Wholesale Corporation (COST) Motorola Solutions Inc (MSI) Sora (SORA) QUALCOMM, Inc. (QCOM)

Top Social Posts

Top posts by engagements in the last [--] hours

"@DavidBCollum @DavidSchawel That's an interesting way to frame a total return swap question"
X Link 2018-10-03T23:31Z [----] followers, [--] engagements

"Scientist stoking fear spreading outdated numbers without doing any due diligence even after the original author updated his estimates"
X Link 2020-01-25T15:45Z [----] followers, [--] engagements

"@ETH_Zurich3773 Intel Capital is the biggest dumb money in SV. Its really good in providing golden parachute to friends and families. Capital incinerator and destroyer of shareholder capital and employee morale"
X Link 2023-01-11T03:56Z [----] followers, [--] engagements

"Reminds me of how Microsoft messed up the Gap Inc earnings release summarization QA demo for Bing chat few months ago. Guys maybe check the SQL query before doing the demo video #MicrosoftBuild #MicrosoftFabric https://t.co/OuLUk8YLep Guys maybe check the SQL query before doing the demo video #MicrosoftBuild #MicrosoftFabric https://t.co/OuLUk8YLep"
X Link 2023-05-24T06:21Z [----] followers, [---] engagements

"@adropboxspace Will take shit tier EB-5 over fast causal Asian American chipotle wannabe food without soul any day"
X Link 2023-05-29T19:41Z [----] followers, [--] engagements

"Too many at $INTC talks product and strategy focusing on growing headcounts instead of doing engineering. We didnt get into this mud hole because everything was going great." Inside Intel's efforts to turn itself around. via @WSJ https://www.wsj.com/articles/intel-gelsinger-nvidia-turnaround-30febac6st=73hrd0qr723rw6j https://www.wsj.com/articles/intel-gelsinger-nvidia-turnaround-30febac6st=73hrd0qr723rw6j"
X Link 2023-05-30T15:56Z [----] followers, [---] engagements

"Aspiring web devs and engineers wishing to get into software: Learn to build your own chips. CMOS flip flops clock trees write your own Verilog implement a small FPGA and make it do something"
X Link 2023-07-09T01:39Z [---] followers, [---] engagements

"@mckaywrigley Langchain focused on on providing abstraction for the latest fab app with [--] lines of code. Great for the wow effect in Twitter demos. But for actual production usage its a huge leaky abstraction problem. Disaster for dev exp. We need langchain core and move demoware out"
X Link 2023-07-09T03:01Z [---] followers, [---] engagements

"@jobergum @altryne @jeffreyhuber Its at best as useful as a random embedding models. No good. Embeddings is cheap to run on CPU anyway"
X Link 2023-07-15T07:09Z [---] followers, [---] engagements

"@deliprao Langchain is great. There are some obvious deficiencies and different factors at play e.g. between DevEx and business models. But thats part of the growth pain. The latest few updates are good small steps in moving towards a more matured direction"
X Link 2023-08-07T18:17Z [----] followers, [---] engagements

"@suchenzang Half of the code base is still done using to be deprecated OpenAI text completion API not yet transitioned into chat completion. Default prompts were from text-davinci-002 days. Oh and theres like [--] layers of abstraction on the common tasks scattered everywhere"
X Link 2023-08-13T04:44Z [---] followers, [--] engagements

"@jxnlco Pretty. patch() is kinda too implicit for users How about patch_openai_package() or something more explicit"
X Link 2023-08-17T04:47Z [--] followers, [---] engagements

"@jxnlco that sounds great as well im assuming that would be a patched openai library that's almost interface parity"
X Link 2023-08-17T05:10Z [---] followers, [--] engagements

"@_onionesque Why buy the cow when you can get milk from supermarkets for a few bucks And where does full stack end"
X Link 2023-08-25T13:57Z [---] followers, [----] engagements

"@Teknium1 @Shahules786 It's data leakage where the features of the HumanEval benchmarks leaked through iterative probing from the Evol-Instruct training data generation process. HumanEval score for WizardLM is not a good metric for its generalized coding ability"
X Link 2023-08-27T05:47Z [---] followers, [--] engagements

"@AviSchiffmann That would be a South Korean VC fund. Or Wood League NA Pleb Fund I"
X Link 2023-08-31T22:45Z [---] followers, [--] engagements

"@snoopy_dot_jpg This is a generational. Carmack is Gen X. Gen X is built different"
X Link 2023-09-02T15:45Z [---] followers, [---] engagements

"@jxnlco @athyuttamre One of the hurdle is that some APIs would have a gigantic thousands of lines OpenAPI spec which makes better definition and auditing difficult. e.g. OpenAIs API spec is [----] lines long. Currently just extracting the relevant pieces out but theres probably a better way"
X Link 2023-09-04T01:56Z [----] followers, [---] engagements

"2. Only Lee Seok-bae can create Lk99 Free range small batch certified organic hand-crafted artisanal #LK99 made with care and love at Korea University. ngmi"
X Link 2023-09-05T03:26Z [---] followers, [---] engagements

"@thisischaniece Well it is the month of ghosts in lunar calendars"
X Link 2023-09-06T04:31Z [----] followers, [---] engagements

"@deliprao Theres least one in there that has nothing to do with AI research engineering or product"
X Link 2023-09-07T13:40Z [---] followers, [----] engagements

"You cannot eat ARR"
X Link 2023-09-21T15:36Z [---] followers, [---] engagements

"@MistralAI This is open source"
X Link 2023-09-27T05:20Z [----] followers, 14.7K engagements

"If you are a product manager you are wasting company resources"
X Link 2023-09-28T06:03Z [---] followers, [---] engagements

"Prompt engineering tricks. lacing chain-of-whatever or fine-tuning on test sets do not fix LLM's deficiencies or inability to reason and plan. They are engineering hacks for building LLM products. Not what should be accepted as okay"
X Link 2023-09-30T19:11Z [---] followers, 17.9K engagements

"@quantian1 Still need [--] more papers published at top tier conferences after solving Reimann Hypothesis be qualified to do ML research at NVidia"
X Link 2023-10-01T02:38Z [---] followers, [----] engagements

"The fallacy here is called self driving car phenomenon; its always [--] years away. We dont know if reasoning will be the next emerging ability. What we do know is its current limitations that we need to engineer around. Dont deny they exist"
X Link 2023-10-01T04:33Z [---] followers, [---] engagements

"@rajiinio Because now everything in context and grounded based on facts has to compete against the guardrails fine tuned in the LLMs. Guardrails became the source of hallucinations"
X Link 2023-10-01T17:33Z [---] followers, [--] engagements

"@8teAPi We have exactly the same problem as AI ethicsists are doing the same"
X Link 2023-10-02T16:50Z [----] followers, [---] engagements

"@nearcyan @POTUS266 outsource investments free up bandwidth"
X Link 2023-10-12T22:46Z [---] followers, [---] engagements

"@jxnlco @LangChain Yup. That's what we've been doing for the past several months. The key question is the level of abstraction i.e. app composite or individual chains and the communication protocol i.e. response streaming or WebSockets"
X Link 2023-10-13T15:41Z [---] followers, [---] engagements

"Embeddings powering search and recommendation since mid 2010s. Whats old is new again in the world of LLMs"
X Link 2023-10-15T22:42Z [---] followers, [---] engagements

"@main_horse @arankomatsuzaki Not too involved with images. But we do not need the writings of all of the books to make an LLM great"
X Link 2023-10-27T09:18Z [---] followers, [--] engagements

"@agihippo Theres limitations to physical layout bus bandwidth and heat profile that is not present in ASICs like TPU"
X Link 2023-10-29T04:46Z [---] followers, [---] engagements

"GPTs is pretty sick. With RAG KB built in"
X Link 2023-11-06T18:31Z [---] followers, [---] engagements

"@Molson_Hart Use GenAI"
X Link 2023-11-09T13:40Z [---] followers, [---] engagements

"@ivanleomk @jxnlco Yeah context length is a huge problem for txt2sql. The idea is running txt2sql using OpenAI API prompt abstraction that their model is trained to do instead of using off the shelf prompts from Langchain/LlamaIndex. Guess we could test it on spider"
X Link 2023-11-17T00:59Z [---] followers, [--] engagements

"@atroyn @typedfemale Was Microsoft a donor or an investor to OpenAI holdco"
X Link 2023-11-20T01:06Z [---] followers, [---] engagements

"New CEO of OpenAI"
X Link 2023-11-20T05:30Z [---] followers, [---] engagements

"@stratechery @jeremyphoward led by a new CEO in Shear Emmett Shear"
X Link 2023-11-20T19:04Z [---] followers, [---] engagements

"@eyfriis @LangChainAI Got it. Was looking at the source code and noted the significant diff between master and 0.0.399 with a lot of things moved to langchain_core. Good to know there are no planned API changes. Downside is that there will be a gap in import/style between current vs planned changes"
X Link 2023-11-21T00:41Z [---] followers, [--] engagements

"Reminder: OpenAI (the non-profit) might still have the option to go nuclear and opensource everything"
X Link 2023-11-21T06:21Z [---] followers, [---] engagements

"@sunilmallya Too late. Its all ASML and TSMC now"
X Link 2023-11-22T07:31Z [---] followers, [--] engagements

"Why would anyone believe that employee #30000+ should have higher net worth than employee #30 and 8% owner at IPO"
X Link 2023-11-25T03:36Z [---] followers, [---] engagements

"@paulg Falcon was a collaboration outsourced to LightOn which is French. There are options to outsource to US based white box LLM companies like Mosaic/Databricks TogetherComputer etc"
X Link 2023-11-25T09:35Z [---] followers, [---] engagements

"Summary of the thread"
X Link 2023-11-27T05:13Z [---] followers, [---] engagements

"Interests in productivity apps dip during the holiday season. ChatGPT Microsoft Excel Salesforce etc all exhibit the same phenomenon"
X Link 2023-12-01T02:26Z [---] followers, [---] engagements

"This is called "alignment" and "AI ethics" in the research literature. Except we "align" for more than one party"
X Link 2023-12-01T04:50Z [---] followers, [---] engagements

"@jobergum trailing edge gpu should suffice. intel/amd will blas their way through this segment"
X Link 2023-12-01T08:05Z [---] followers, [--] engagements

"Quantum computing is pure research. It will be practical when news came out of China South Korea or Taiwan"
X Link 2023-12-03T23:30Z [---] followers, [---] engagements

"Old enough to remember OpenAI Gym and OpenAI Spinup when it was open. And now Meta is open-er"
X Link 2023-12-09T05:02Z [---] followers, [---] engagements

"This. So much. This is not the first time e/acc open access (closed data) folks bullying academics for not including their models for evaluations"
X Link 2023-12-09T22:49Z [---] followers, [---] engagements

"@TheXeophon Anyone who has ever worked in finance knows no stats/ML/LLM can do better than equity analysts in forecasting the next quarter. Ever"
X Link 2023-12-10T02:54Z [---] followers, [--] engagements

"@simonw Azure release so far no"
X Link 2023-12-13T03:59Z [---] followers, [----] engagements

"OpenAI changing terms to prohibit distillation"
X Link 2023-12-13T07:57Z [---] followers, [---] engagements

"As an example Gemini API has a field named tools aligning with OpenAI API but it has very different content fields structure. It is not drop in compatible with OpenAI SDK so users would need to rely on orchestrators like @LangChainAI or build API switches"
X Link 2023-12-15T17:47Z [---] followers, [--] engagements

"@jxnlco @LangChainAI TBF @openai discontinued publishing ChatML so finetuning data prompt formatting is devolving. Integration cost of a new OSS or open weight model will be sky high if is not supported by Axolotl vLLM or llama.cpp. Dont turn LLMs into Linux versions"
X Link 2023-12-15T18:36Z [---] followers, [--] engagements

"Historically Intel was always able to salvage design decisions with its manufacturing dominance. They lost the manufacturing edge by over incentivizing product and marketing. TSMs execution is a major factor in AMD and NVIDIAs success. I worked at Intel on Larrabee applications in [----]. Then I went to NVIDIA to work on ML in [----]. So I was there at both places at that time and I can say: NVIDIA's dominance didn't come from luck. It came from vision and execution. Which Intel lacked. I worked at Intel on Larrabee applications in [----]. Then I went to NVIDIA to work on ML in [----]. So I was there"
X Link 2023-12-20T20:43Z [---] followers, [---] engagements

"@abacaj Could be. At the same time Mixtral might not be as strong in common use cases. At the very least it doesn't know as many languages as Claude Cohere Gemini and OpenAI. Mistral 7B Dolphin hallucinates way less than Mixtral for some of my use cases"
X Link 2023-12-21T06:05Z [---] followers, [---] engagements

"To be fair VCs invent their own benchmarks as well. Until they run into institutionals that tell them You cant eat IRR. Are academic benchmarks for academics or for investors Are academic benchmarks for academics or for investors"
X Link 2023-12-21T21:59Z [---] followers, [---] engagements

"Back in [----] I put out a research note (sell side not academic). Title: Lies damn lies and AMDs benchmarks. First party benchmarks are for marketing. Paid 3rd party benchmarks are marketing spend. Welcome to tech"
X Link 2023-12-22T08:02Z [---] followers, [---] engagements

"@lateinteraction @mbusigin Would love to see interoperability with the @langchain ecosystem"
X Link 2023-12-22T19:58Z [---] followers, [---] engagements

"The obsession with shoving LLM to solve coding problems and automation is at best short-sighted. Dev Tooling TAM is at best only 2x Nvidia's sales. Dream bigger"
X Link 2023-12-22T20:14Z [---] followers, [---] engagements

"@lateinteraction @hwchase17 @mbusigin @LangChain The ability to show/export the full prompt including default/seed instructions examples and demos to LLM orchestration layers"
X Link 2023-12-23T04:46Z [---] followers, [---] engagements

"Bytedance and Mihoyo are laughing their way to the bank. this is what research papers look in the current year of our lord. looks like japan has had the last laugh https://t.co/2jem8MJaMO this is what research papers look in the current year of our lord. looks like japan has had the last laugh https://t.co/2jem8MJaMO"
X Link 2023-12-25T07:59Z [--] followers, [---] engagements

"When FANG was coined Apple Microsoft and Nvidia were in a decade long depressed valuation multiple regimes. Netflix was the darling in cord cutting transition from cable to streaming after they navigated the [----] fucked up. how did Netflix sneak into FAANG and yet Microsoft got excluded. how did Netflix sneak into FAANG and yet Microsoft got excluded"
X Link 2023-12-27T18:07Z [---] followers, [---] engagements

"Absent of memorization it would be easier to win the lottery than to generate already written text verbatim. What happens if lets say OAI is forced to retrain without NYT articles (won't happen but let's say they do) - and it still outputs verbatim NYT articles because they simply spit out most likely next tokens in their own work What then Do we ban AI entirely lol It is more What happens if lets say OAI is forced to retrain without NYT articles (won't happen but let's say they do) - and it still outputs verbatim NYT articles because they simply spit out most likely next tokens in their"
X Link 2023-12-27T20:42Z [--] followers, [----] engagements

"I read* all [--] 📃pages of the New York Times vs Microsoft and OpenAI complaint. The implications are 🔥stunning**. A 🧵: * with ChatGPT file upload. ** generated by ChatGPT"
X Link 2023-12-28T01:16Z [---] followers, [---] engagements

"Current LLMs are lossy compressions. LLM can also be trained to do lossless compressions. Text embeddings reveal as much as text. NYT/OpenAI lawsuit completely misunderstands how LLMs work and judges getting this wrong will do huge damage to AI. Basic point: LLMs DON'T "STORE" UNDERLYING TRAINING TEXT. It is impossible- the parameter size of GPT-3.5 or [--] is not enough to losslessly encode the training set. NYT/OpenAI lawsuit completely misunderstands how LLMs work and judges getting this wrong will do huge damage to AI. Basic point: LLMs DON'T "STORE" UNDERLYING TRAINING TEXT. It is"
X Link 2023-12-28T02:40Z [---] followers, [----] engagements

"The best way to tell your customers you hate them is by outsourcing support. The second best way is to build a chatbot"
X Link 2023-12-28T17:14Z [--] followers, [---] engagements

"@abacaj Are we seriously suggesting that those blank check companies need to use GPU poor techniques to improve their models Especially when majority of these TPOT open source models are results of distilling OpenAIs models"
X Link 2023-12-28T23:30Z [----] followers, [---] engagements

"This is the equivalent of looting artifacts placing them in the British Museum and claiming them as British. this is how you win https://t.co/bbV9reuvrK https://t.co/EN0mTQ5z4A this is how you win https://t.co/bbV9reuvrK https://t.co/EN0mTQ5z4A"
X Link 2024-01-03T23:28Z [---] followers, [---] engagements

"@mueblesfeos AFAIK the conflict between companies vs shareholders seeking liquidity is the reason some of the earlier incarnations of private markets failed. The good old days of getting private shares off secondary market from divorces or plane crashes"
X Link 2024-01-08T18:13Z [--] followers, [--] engagements

"Not sure what's the point of recycling old frameworks and acronyms with slightly different spelling. Both augmented intelligence and intelligence amplification have been around for decades. Also from a [----] paper: ChatGPT is level 3"
X Link 2024-01-08T23:16Z [---] followers, [---] engagements

"It does feel like self-driving cars all over again with no clear definitions on the levels until [----]. FFA on all the claims and every model/service are AI AGI SGI or Doraemon"
X Link 2024-01-08T23:22Z [--] followers, [---] engagements

"Need to be more Jensen than Jensen. The man himself does not wear the same clothes every day @ $1.3T market cap. Asking for a friend but at what market cap can you pull a Jensen Huang and wear the same clothes every day Massive alpha in not having to care. Asking for a friend but at what market cap can you pull a Jensen Huang and wear the same clothes every day Massive alpha in not having to care"
X Link 2024-01-11T04:15Z [---] followers, [---] engagements

"This video shows the disconnect between NVIDIA and ML engineers. Installing CUDA on WSL is not hard despite poor docs. The real challenge is finding the matching driver version CUDA version cudnn version and the matching PyTorch version so it trains"
X Link 2024-01-13T02:03Z [---] followers, [----] engagements

"Old enough to remember cudnn broke LSTM on both Tensorflow and PyTorch in a few network configurations from their "optimizations". Any architecture not in NVIDIA's optimization path is at risk"
X Link 2024-01-13T02:09Z [---] followers, [---] engagements

"@alignment_lab It's NVIDIA's driver problem. Windows drivers got more frequent updates due to other utilities. Linux drivers don't get frequent updates thus less breakage there. Lambda stack is great if the rig is not used for anything else. But it can't run Crysis"
X Link 2024-01-13T08:13Z [---] followers, [--] engagements

"LLM model generations in OpenAI. 4th Gen: GPT-4 Claude [--] Gemini Ultra() Mixtral Large() 3.5th Gen: GPT-3.5 Claude Mixtral Medium Gemini Pro 3rd Gen: Llama [--] Mixtral 8x7B etc"
X Link 2024-01-13T20:01Z [---] followers, [--] engagements

"This actually says more about the business model and age of different companies. 10% of Apple or Amazon sales is an astronomical number by all standards. Meta spends twice more on R&D than Amazon Alphabet and Microsoft and [--] times more than Apple when normalized by revenue. The only other big tech that comes close is Nvidia. That tells you a lot about who is building the technology of tomorrow. Meta spends twice more on R&D than Amazon Alphabet and Microsoft and [--] times more than Apple when normalized by revenue. The only other big tech that comes close is Nvidia. That tells you a lot about"
X Link 2024-01-14T21:31Z [---] followers, [---] engagements

"Did you know that the NSA implemented TEMPEST certification w.r.t. compromising radiations from communications and comsec equipment in the 1960s Did you know Wim van Eck demonstrated CRT phreaking in the 1980s did you know that researchers in [----] demonstrated that you could process an opponent's microphone extracting the noise induced by their GPU and use that noise to know where they were on a Counter-Strike map https://t.co/k4pfzXvTtZ did you know that researchers in [----] demonstrated that you could process an opponent's microphone extracting the noise induced by their GPU and use that"
X Link 2024-01-17T06:54Z [---] followers, [---] engagements

"Pythia Bloom LLM360 are all open weight open source and open data lineage. @amasad 💯. The Linux of LLMs does not yet exist. @amasad 💯. The Linux of LLMs does not yet exist"
X Link 2024-01-18T08:23Z [---] followers, [---] engagements

"Microsoft under Ballmer runs the worlds second largest social network invested in Facebook acquired Skype and Yammer. It also built a gaming business. It tried Windows Phone but no one won vs Motorola and Nokia until iPhone which was a Cisco trademark. @felipeclopes @GergelyOrosz @petergyang Is this true Pre-Nadella Microsoft was nowhere on the most important computing paradigms (phones & social). If it wasn't for bundling Office365 where would it be @felipeclopes @GergelyOrosz @petergyang Is this true Pre-Nadella Microsoft was nowhere on the most important computing paradigms (phones &"
X Link 2024-01-18T08:36Z [---] followers, [---] engagements

"350K Nvidia H100. 250K AMD. ASICs improbable unless they secure wafer starts last year"
X Link 2024-01-18T21:58Z [--] followers, [--] engagements

"@YunTaTsai1 @surmenok If AMD Intel and TI couldnt sustain its IDM model with their volume doubt if anyone else can. High fix cost manufacturing business with specialized talents"
X Link 2024-01-20T01:36Z [---] followers, [--] engagements

""Ninety percent of success in life is just showing up." - Woody Allen"
X Link 2024-01-25T06:44Z [---] followers, [---] engagements

"@abacaj @Weyaxi TBF this holds true for any proprietary models behind an API facade. At that point the evaluation is on the ML system instead of teh ML model itself. An example Azure OpenAi has guardrails enabled by default which means 5% of inputs will get randomly censored for no reason"
X Link 2024-01-28T01:49Z [---] followers, [---] engagements

"China will not catch up to Taiwan South Korea or even the U.S. for semiconductor manufacturing. It will leap-frog the incumbents on new substitute technology as it did in wireless solar LED HSR consumer tech drones AI/ML and soon electric vehicles. https://t.co/k5EDePFb2u Okay I wrote so much about this last year and its nice to see the man himself say yeah BYD Chinas budding semiconductor (and car) company will murder everyone in the world. Its completely true. https://t.co/k5EDePFb2u Okay I wrote so much about this last year and its nice to see the man himself say yeah BYD Chinas budding"
X Link 2024-01-28T08:12Z [---] followers, [---] engagements

"These secret question weights are literally: Let's play the "Get hired as an Air Traffic Controller" game. 90% of you will fail. Are you ready The first question is an easy one if you've been paying attention: https://t.co/LKSEMpuRoq Let's play the "Get hired as an Air Traffic Controller" game. 90% of you will fail. Are you ready The first question is an easy one if you've been paying attention: https://t.co/LKSEMpuRoq"
X Link 2024-01-30T21:00Z [---] followers, [---] engagements

"@TShirtnJeans2 @dylan522p it depends. friendly sell-sides usually get callbacks since they represent clients/buy-sides. unfriendly sell-side could get a hard time. calibrate and align expectations. this is specific to callbacks after earnings call. in the old days"
X Link 2024-02-01T08:09Z [---] followers, [--] engagements

"Hmm. So RAGAS defaults to use dot similarity not cosine Not all encoder outputs are normed like word2vec or openai's embeddings"
X Link 2024-02-02T04:39Z [---] followers, [---] engagements

"image metadata self-contained or generated is way more effective than image search. There's a reason why social media uses hashtags. Self-supervised feature tagging. Embedding vectors is a feature not a complete representation. @felix_red_panda it's pretty good for image search imo. but text vectors definitely suffer from this problem @felix_red_panda it's pretty good for image search imo. but text vectors definitely suffer from this problem"
X Link 2024-02-02T07:22Z [---] followers, [---] engagements

"@lateinteraction Langchain. Or to lessor extend LlamaIndex. Would stay away from proxy abstractions that routes LLM calls like @LiteLLM or @withmartian or @PortkeyAI. They all provide pseudo OpenAI compatible interface to solve users problems. But does not solve framework problems"
X Link 2024-02-03T20:16Z [---] followers, [---] engagements

"@lateinteraction @LiteLLM @withmartian @PortkeyAI Do we really need to solve #2 though Thats usually handled at the model serving API layer. So we send async calls via direct httpx API calls vender SDKs or Langchain. Maybe locally to Ollama. Integration game is a distraction"
X Link 2024-02-03T20:34Z [---] followers, [---] engagements

"@lauradietz99 Yes makes sense for encoders like OpenAI's embeddings that are already normed but break down for other embeddings. Most "AI Engineers" don't have a clue about this which results in misleading evals"
X Link 2024-02-04T06:54Z [---] followers, [--] engagements

"@abacaj ChatGPT excels at answers that can be found on Stackoverflow. It's hallucination-central when there's no direct answer but only half-complete responses in disjoint Github issues and Stackoverflow comments"
X Link 2024-02-06T05:38Z [---] followers, [----] engagements

"@jxnlco inject functions into functions. booya. def indexer(encoder: Callable = openai_embed) - bool:"
X Link 2024-02-06T19:05Z [---] followers, [---] engagements

"Why are people that curious about ChatGPT system prompt"
X Link 2024-02-07T15:02Z [---] followers, [----] engagements

"1. This is trading time complexity for space complexity. Good trade. [--]. NN is awesome at connecting the dots. In language it would be hallucination. Google Deepmind presents Grandmaster-Level Chess Without Search paper page: https://t.co/Drw8kduoYH largest model reaches a Lichess blitz Elo of [----] against humans and successfully solves a series of challenging chess puzzles without any domain-specific tweaks or explicit https://t.co/vbqUockfqQ Google Deepmind presents Grandmaster-Level Chess Without Search paper page: https://t.co/Drw8kduoYH largest model reaches a Lichess blitz Elo of 2895"
X Link 2024-02-08T15:01Z [---] followers, [---] engagements

"AWS Bedrock and GCP Vertex AI sucks. Its a competition on how to make DevEx horrid. Self hosting using vLLM is much more pleasant"
X Link 2024-02-10T21:29Z [---] followers, [---] engagements

"There's open source and there's open Source as a fundraising/marketing strategy. They are not the same. Skiff raised $14m in VC to build an open source alternative Google Suite. Now it's being acquired by Notion and the code has been taken down. All software is ephemeral. File over app. https://t.co/pS47oOTUBz Skiff raised $14m in VC to build an open source alternative Google Suite. Now it's being acquired by Notion and the code has been taken down. All software is ephemeral. File over app. https://t.co/pS47oOTUBz"
X Link 2024-02-11T00:21Z [---] followers, [---] engagements

"you must be a peasant if you questioned technology that was about to be adopted in such a Serious Enterprise full of Serious People. After all it was being built by Accenture no less (Cue eye-roll.)"
X Link 2024-02-12T06:38Z [---] followers, [---] engagements

"Santa Clara 49ers"
X Link 2024-02-13T01:22Z [---] followers, [---] engagements

"Lots of hindsight on NVIDIA and plenty of clickbaits. NVIDIA fought tooth and nail vs Intel ATI/AMD across many major technology shifts including consoles chipsets mobiles etc. Got their ass kicked in some. GPGPU from [----] wasn't a real thing until crypto and now AI"
X Link 2024-02-13T05:38Z [---] followers, [---] engagements

"The swarm of papers claiming they are the first to use ChatGPT or GPT for NLG evaluation in [----] is interesting"
X Link 2024-02-14T06:26Z [---] followers, [---] engagements

"@bernhardsson Pool together reserved GPU instances securitization vehicles and offload the risk to some fintech traders. It's been a while since Enron or crypto bust"
X Link 2024-02-14T22:12Z [---] followers, [---] engagements

"OpenAI takes on Google Microsoft Baidu Yandex Yahoo DuckDuckGo Perplexity You by developing its own web search product. New: OpenAI wants to take on Google by developing its own web search product https://t.co/YuHG12k4bo New: OpenAI wants to take on Google by developing its own web search product https://t.co/YuHG12k4bo"
X Link 2024-02-14T22:21Z [---] followers, [---] engagements

"This is hella fast. Missing UX: [--]. Support for pyproject.toml is not documented. [--]. No install groups. [--]. Uses venv to enable virtual env. Poetry uses itself. [--]. Listing packages in venv require pip. (same as poetry) [--]. No lock files. [--]. No scripting using pyproject Announcing uv: an extremely fast Python package installer and resolver written in Rust. uv is designed as a drop-in alternative to pip pip-tools and virtualenv. With a warm cache uv installs are near-instant. Here it's 75x faster than pip and pip-tools. https://t.co/wrvaudUn6i Announcing uv: an extremely fast Python package"
X Link 2024-02-15T21:35Z [---] followers, [----] engagements

"Still early very raw. pip replacement but not feature parity with poetry yet"
X Link 2024-02-15T21:39Z [---] followers, [---] engagements

"@charliermarsh Ohh it does. Thanks Was used to pip list. uv pip freeze grep "one of the package in pyproject.toml" one package is missing from pyproject.toml so prob no support yet"
X Link 2024-02-15T21:43Z [---] followers, [---] engagements

"@charliermarsh Also which uv it runs under a directory/venv is not clear. usually if venv will use the pip inside that venv. for uv it uses system pip"
X Link 2024-02-15T21:44Z [---] followers, [---] engagements

"Seems some of the source data might involve NeRFing midjourney generations I ran all of the Sora prompts through Midjourney Interesting how similar some are side-by-sides against vids: I ran all of the Sora prompts through Midjourney Interesting how similar some are side-by-sides against vids:"
X Link 2024-02-16T14:29Z [---] followers, [---] engagements

"ai aoi sora when"
X Link 2024-02-16T21:18Z [---] followers, [---] engagements

"OpenAI acquired Global Illumination an OSS MMORPG company. Large-scale high-quality training data doesn't come out of exhausting token limits. This is equivalent to learning how to drive from running simulations in GTA several years back. I'm so curious about the details of Sora. Are there fundamentally new ideas here or is it just unimaginable scale of data and compute I remember when DALL-E came out the open-source equivalents were at least comparable. Buthere it's not even close. https://t.co/XBTSjeLkcP I'm so curious about the details of Sora. Are there fundamentally new ideas here or is"
X Link 2024-02-17T05:39Z [---] followers, [---] engagements

"Grand Theft Auto V also known as synthetic data In 2024"
X Link 2024-02-17T06:44Z [---] followers, [---] engagements

"Sora is really something else"
X Link 2024-02-17T06:52Z [---] followers, [---] engagements

"Again NIAH is a naive evaluation. Recall drops to 60% when there are [---] "needles". This is all before long-form question answering that might involve logic and reasoning. oh my goodhart https://t.co/l3loCgA1AU oh my goodhart https://t.co/l3loCgA1AU"
X Link 2024-02-17T08:17Z [---] followers, 20.3K engagements

"@Francis_YAO_ I disagree. This is a misunderstanding of memory hierarchy architecture. Running a 24TB DRAM instance does not render AWS S3 unnecessary. For EECS it's a cache hierarchy design. For CS it's data management design. For IR it's a precision/recall tradeoff"
X Link 2024-02-17T19:32Z [---] followers, 12.9K engagements

"@yacineMTB wheres dingboard lebron feet"
X Link 2024-02-20T03:54Z [---] followers, [---] engagements

"@atroyn It is a difference in design philosophy between LLM as a computer and LLM as a system. Currently most models are trained as a system with their own biases"
X Link 2024-02-23T19:44Z [---] followers, [---] engagements

"@karpathy @debarghya_das Any tips and tricks on using Obsidian Theres a learning curve to it and not a whole lot of resources online"
X Link 2024-02-24T17:31Z [---] followers, [----] engagements

"@TheXeophon Netflix prize resulted in a few quant shops and most importantly Apache Spark and Databricks"
X Link 2024-02-24T18:06Z [---] followers, [---] engagements

"@eugeneyan Theres not a very good source of information to bridge the gap between configuring note apps vs taking notes. There are ample walkthroughs online about configuring vim setting up vscode building your own rig etc. Not for note taking. All I have atm are MDs"
X Link 2024-02-24T18:12Z [---] followers, [---] engagements

"@gSajko @eugeneyan Was using Notepads and Apple Notes only. But age has taken its toll and I need all the help I can get"
X Link 2024-02-24T20:00Z [---] followers, [--] engagements

"Count the number of open software engineering positions at Nvidia. Watch not what they say but what they do. Jensen Huang CEO of Nvidia argues that we should stop saying kids should learn to code. He argues the rise of AI means we can replace programming languages with human language prompts thus enabling everyone to be a programmer. AI will kill coding. https://t.co/SxK9twhEby Jensen Huang CEO of Nvidia argues that we should stop saying kids should learn to code. He argues the rise of AI means we can replace programming languages with human language prompts thus enabling everyone to be a"
X Link 2024-02-26T00:34Z [---] followers, [---] engagements

"Not only the generated facial muscle movements matches vocal intonations its able to generate different facial muscle movements styles as well. https://t.co/8C2gyReiSX https://t.co/8C2gyReiSX"
X Link 2024-02-28T05:18Z [---] followers, [---] engagements

"@HamelHusain @_hex_tech How does that handle virtual environments Jupyter ecosystem is so tied down to conda its hard for DS to bridge between their notebooks and prod env"
X Link 2024-02-29T07:41Z [---] followers, [---] engagements

"@sbeastwindy Google AI Studio"
X Link 2024-03-01T07:35Z [---] followers, [--] engagements

"This is how Singapore Taiwan South Korea and Hong Kong industrialized. Single party authoritarian rules. great quote from karpathy most great organizations require leader(s) with a disproportionate amount of power when this is absent you end up with countless hierarchies of ineffective committees e.g. many google products lack a Directly Responsible Individual with actual power https://t.co/esiO7X9t94 great quote from karpathy most great organizations require leader(s) with a disproportionate amount of power when this is absent you end up with countless hierarchies of ineffective committees"
X Link 2024-03-01T08:14Z [---] followers, [---] engagements

"Elon Musk sues OpenAI over breach of fiduciary duty"
X Link 2024-03-01T08:16Z [---] followers, [---] engagements

"@FoolAllTheTime Tick tock was a strategy centered around manufacturing not around design. Nvidia isnt idm and cant replicate the same. Also old enough to remember yield issues with Ampere"
X Link 2024-03-03T07:02Z [---] followers, [---] engagements

"TBF Emma Stone played an Asian character. I'm seeing some talk about Robert Downey Jr ignoring Ke Huy Quan but I'm not seeing anything about Emma Stone doing basically the same thing to Michelle Yeoh https://t.co/eh97F56Q61 I'm seeing some talk about Robert Downey Jr ignoring Ke Huy Quan but I'm not seeing anything about Emma Stone doing basically the same thing to Michelle Yeoh https://t.co/eh97F56Q61"
X Link 2024-03-11T17:31Z [---] followers, [---] engagements

"Electronic Design Automation (EDA) revolutionized semiconductor design in the 80s and 90s expanding the pool of design engineers as it grew. Software Design Automation tools are set to boost the number of software engineers. SDA hurts WITCH and non-tech SDEs. This reminds me a lot of when Napster got big in [----] It took the music industry fighting MP3 file sharing for a decade before embracing it and coming up with a digital alternative to CDs which was Spotify (and yes it's owned by the music industry) Here we see the same: This reminds me a lot of when Napster got big in [----] It took the"
X Link 2024-03-12T22:23Z [---] followers, [---] engagements

"@hwchase17 Copy-pasta fix in the import statement block is fine. Logic changes and new dependencies are high friction"
X Link 2024-03-14T19:46Z [---] followers, [---] engagements

"Cool study. But its against TOC to use OpenAI for employment decisions. 🚨Read this before you use OpenAI for Hiring🚨 @LeonYin @daveyalba and I ran thousands of resume screening tests with GPT-3.5 and GPT-4 and found that the tech will racially discriminate applicants based only on their name. Serious implications. 🧵 1/10 https://t.co/XDcngpq5pa https://t.co/u5IwBVuHyU 🚨Read this before you use OpenAI for Hiring🚨 @LeonYin @daveyalba and I ran thousands of resume screening tests with GPT-3.5 and GPT-4 and found that the tech will racially discriminate applicants based only on their name."
X Link 2024-03-15T06:17Z [---] followers, [----] engagements

"@sbeastwindy @elonmusk Gotta respect any men that keep his words"
X Link 2024-03-17T21:16Z [---] followers, [--] engagements

"Instability Diffusion https://sifted.eu/articles/stability-ai-rombach-news https://sifted.eu/articles/stability-ai-rombach-news"
X Link 2024-03-20T23:49Z [----] followers, [---] engagements

"@jobergum thats a very good suggestion. ill get devin to work on this"
X Link 2024-03-21T07:20Z [---] followers, [---] engagements

"@jobergum With vector databases such as pinecone organizations will gain access to a broad set of capabilities they need to build GenAI applications with security privacy and responsible AI"
X Link 2024-03-21T07:38Z [---] followers, [---] engagements

"The synergy of tech hubs network and compounding improvements in crafts means US-based software engineers exponentially more adept than elsewhere. Exponentially"
X Link 2024-03-23T06:04Z [---] followers, [---] engagements

"@ptsi Theres always Salesforce the software vs sales force the people"
X Link 2024-03-23T17:18Z [---] followers, [--] engagements

"given the title inflation of recent years data scientist is now person who has no engineering skills and whose work has no scientific rigor. Data scientists are a different breed man They'll write the most fucked up Python you've ever seen and with a straight face ask "Does this look right" I have seen Jupiter notebooks that make the Necronomicon look like "The Very Hungry Caterpillar" Data scientists are a different breed man They'll write the most fucked up Python you've ever seen and with a straight face ask "Does this look right" I have seen Jupiter notebooks that make the Necronomicon"
X Link 2024-03-28T14:55Z [---] followers, [---] engagements

"even alibaba has their open weight model. wheres aws Alibaba Qwen release Qwen1.5-MoE-A2.7B. MoE model with 2.7B activated parameters but it can achieve the 7B model performance in a series of benchmark evaluation. Additionally compared with the 7B model it has significant advantages in training costs and inference speed. https://t.co/1S3eXxBgkk Alibaba Qwen release Qwen1.5-MoE-A2.7B. MoE model with 2.7B activated parameters but it can achieve the 7B model performance in a series of benchmark evaluation. Additionally compared with the 7B model it has significant advantages in training costs"
X Link 2024-03-28T22:06Z [---] followers, [---] engagements

"@gazorp5 they helped creating the open source but not really open source licenses at mongo elastic and now redis"
X Link 2024-03-28T22:39Z [---] followers, [--] engagements

"@gazorp5 that's a balance sheet transaction not income statement. but yeah people get paid selling promises"
X Link 2024-03-29T02:24Z [---] followers, [--] engagements

"mckinsey is gonna boeing redhat into ibm. oh w8. https://www.theregister.com/2024/03/27/red_hat_hires_mckinsey/ https://www.theregister.com/2024/03/27/red_hat_hires_mckinsey/"
X Link 2024-04-01T00:09Z [---] followers, [---] engagements

"Irwin Jacobs founder of Qualcomm. His son Paul Jacobs PhD from Cal EECS and later became CEO of Qualcomm. Sutardja and Dai of Marvell. Son went to Cal EECS. can't speak about their private lives. I've been looking for an example of this for a year and I'm pretty sure they don't exist: - Worth 1B+ - Married to original spouse - Outstanding moral virtue - Kids are all relatively successful - Kids have a good relationship with parents Any idea cc @FoundersPodcast I've been looking for an example of this for a year and I'm pretty sure they don't exist: - Worth 1B+ - Married to original spouse -"
X Link 2024-04-02T17:42Z [---] followers, [----] engagements

"AI Engineer as defined by OpenAI's job description: Driven and innovative ML software engineer as our founding AI Engineer to ideate scope and train custom models for previously unsolvable problems that deliver significant value for our customers. https://openai.com/careers/ai-engineer https://openai.com/careers/ai-engineer"
X Link 2024-04-05T16:51Z [---] followers, [----] engagements

"using ticker as model name on huggingface before going public is so alpha. dbrx"
X Link 2024-04-05T22:49Z [---] followers, [---] engagements

"Search/RecSys: data flywheel to @databricks Spark job to retrain models hourly serve real time inference or dump to Redis. LLM Apps: data flywheel to Langsmith prompt tuning via DSPy and persist to Langsmith. @LangChainAI pull latest prompt. NGMI with fine-tuning first. Fine tuning popping up on my timeline again. Ive said it before but I was wasting a lot of time ft open source models only to get marginal returns for products that did not have pmf or strong growth. There is a lot of overhead to ft its not one and done - data changes you Fine tuning popping up on my timeline again. Ive said"
X Link 2024-04-05T23:12Z [---] followers, 30K engagements

"UCLA stands for U C Lots of Asians. No idea where the Los Angeles myth came from. Why can you say UC San Diego but not UC Los Angeles Why can you say UC San Diego but not UC Los Angeles"
X Link 2024-04-06T00:32Z [----] followers, [---] engagements

"@diegoasua @databricks @LangChainAI yes. it very good"
X Link 2024-04-06T13:51Z [---] followers, [---] engagements

"cyclical inbred agi just like European royalties. Model bree merging on HF is getting out of hand https://t.co/dak1trTkSa Model bree merging on HF is getting out of hand https://t.co/dak1trTkSa"
X Link 2024-04-06T22:58Z [---] followers, [---] engagements

"man wtf im getting anthropic snowflake and databricks ads on instagram feeds. leave me alone"
X Link 2024-04-09T06:21Z [---] followers, [---] engagements

"someone start a content farm rick rolling openai crawlers kthx lmao why do people do this https://t.co/58fzaW0AYx lmao why do people do this https://t.co/58fzaW0AYx"
X Link 2024-04-11T18:13Z [---] followers, [---] engagements

"@yacineMTB genai dao crypto conference"
X Link 2024-04-22T21:43Z [---] followers, [---] engagements

"ollama has the best logo/mascot in all these genai companies @yar_vol @martin_casado Ok baited. Will run https://t.co/xCIgyHuMb2 @yar_vol @martin_casado Ok baited. Will run https://t.co/xCIgyHuMb2"
X Link 2024-04-22T21:46Z [---] followers, [---] engagements

"it did. portfolios exits include Tesla Solar City SpaceX Palantir etc. Why does the US government not do venture capital Seems pretty net win. Bet on America reap the upset. Why does the US government not do venture capital Seems pretty net win. Bet on America reap the upset"
X Link 2024-04-24T22:02Z [---] followers, [---] engagements

"real chanel is very expensive. it can cost $3000-$40000 per handbag. that means to outfit for a night with real chanel can cost $10000 to $100000. chinatown knockoff is very cheap. use knock off. no one cares about you. Real surgery is VERY expensive it can be $7500-$15000/hour for a good surgeon That means weekly surgeries with a real doctor can cost you $30000 to $60000 per month Also a real surgeon isn't available 24/7 try calling them at 4am Nights often when you need a surgery the Real surgery is VERY expensive it can be $7500-$15000/hour for a good surgeon That means weekly surgeries"
X Link 2024-04-26T02:41Z [---] followers, [---] engagements

"@ETH_Zurich3773 these small shops are great imo. there are some 100m aum fundamental shops 2-20 returns capital beyond capacity. consistent return. long wait list. the opposite of asset gathering billion dollar seed stage vc fund"
X Link 2024-04-26T06:43Z [----] followers, [--] engagements

"TIL Mexico Costa Rica and Chile work more hours than South Korea. Or the statistics has been dressed for presentation. Maybe a part of the reason the US kicks ass in tech and we suck over here. https://t.co/Is4Pao3QBK Maybe a part of the reason the US kicks ass in tech and we suck over here. https://t.co/Is4Pao3QBK"
X Link 2024-04-26T18:45Z [---] followers, [---] engagements

"takes a lot to say the quite part out loud. lots of shady stuff going on. everyone has vested interests so rarely anyone would call out bullshits. hurts future monetization and employment prospects. easier to pocket marketing budgets than be mkbhd. We need a MKBHD for AI software. Because theres tons of bullshit out there to wade through We need a MKBHD for AI software. Because theres tons of bullshit out there to wade through"
X Link 2024-04-27T01:41Z [---] followers, [---] engagements

"interesting that he mentioned Google not GCP. Mission: make Google the best home for developers building with AI. Mission: make Google the best home for developers building with AI"
X Link 2024-04-27T23:14Z [---] followers, [----] engagements

"@AmgadGamalHasan thats above his pay grade and a perpetual Google problem. Googles chat product has had at least a dozen different names/incarnations"
X Link 2024-04-28T01:49Z [---] followers, [--] engagements

"@docs2info the mbb style of people last management. and the outcome is ge intel and boeing. Google is heading towards that direction but theres still signs of life"
X Link 2024-04-29T00:38Z [---] followers, [--] engagements

"devin and swe agent are much more sophisticated than github copilot where they can provide both the novelty and verify the utlity. and at the end of the day the end user will have to approve the code to grant it the credibility"
X Link 2024-04-30T05:34Z [---] followers, [---] engagements

"influencer posts provides novelty and some credibility by either providing sources or by showing off follower counts. its up to the readers to determine the information's utility. just cos its written doesnt mean its worth shit"
X Link 2024-04-30T05:36Z [---] followers, [---] engagements

"how to determine which javascript framework to learn uwu page w/ sawaratsuki logo"
X Link 2024-05-02T05:11Z [---] followers, [---] engagements

"@hu_yifei also not kosher with naming schema dictated in llama-3 license"
X Link 2024-05-02T14:56Z [---] followers, [--] engagements

"@ankrgyl wait are all bedrock models openai api compatible last i checked we need to do quite a bit of payload manipulations"
X Link 2024-05-03T19:09Z [---] followers, [---] engagements

"@ankrgyl by openai compatible models do you mean openai api compatible inference services also how is the service tracking/allocating tpm/rpms"
X Link 2024-05-03T19:52Z [---] followers, [--] engagements

"weve seen this in the past where general purpose cpu/mcu performs as good as asics and are more future proof. from networking/telco asics to most recently bitcoin mining rigs. general purpose compute always wins. 🦙🦙🦙 [-----] tokens per second 🛫Meta Llama [--] takes off with #TensorRT #LLM 📚➡ https://t.co/A38N2KEy4u @AIatMeta https://t.co/APLQDo9xJH 🦙🦙🦙 [-----] tokens per second 🛫Meta Llama [--] takes off with #TensorRT #LLM 📚➡ https://t.co/A38N2KEy4u @AIatMeta https://t.co/APLQDo9xJH"
X Link 2024-05-07T22:45Z [---] followers, [---] engagements

"this definition of tech industry is interesting. has samsung but no asml nvda tsmc. and wheres nok eric and siemens 1/ Europe doesn't have a tech industry. But why is that the case The short answer: it's not because of regulation. 🧵A thread: https://t.co/wBkpVSiESC 1/ Europe doesn't have a tech industry. But why is that the case The short answer: it's not because of regulation. 🧵A thread: https://t.co/wBkpVSiESC"
X Link 2024-05-08T13:04Z [----] followers, [---] engagements

"we are further and further away from agi Apple has closed in on an agreement with OpenAI to use the startups technology on the iPhone part of a broader push to bring artificial intelligence features to its devices https://t.co/qjmMB3wpsb Apple has closed in on an agreement with OpenAI to use the startups technology on the iPhone part of a broader push to bring artificial intelligence features to its devices https://t.co/qjmMB3wpsb"
X Link 2024-05-11T02:42Z [----] followers, [---] engagements

"@lithos_graphein Samsung is both IDM + foundry not pure play. aka how UMC slowly faded with its relationship with MediaTek"
X Link 2024-05-12T19:42Z [----] followers, [---] engagements

"Section [---] and Section [---] of the Trade Section [---] of the Trade Act of [----] enables US government to place trade measures to protect domestic industries. We placed trade restrictions vs Japan South Korea and Taiwan in the past. China now joins the rank as they industrialize. Oh boy. Even community noters cant read. Hey dummies: Do you know what FREE market means UNRESTRICTED access ZERO government intervention FREE movement of goods Can any American company FREELY export US-made cars to China No. 🙄 Oh boy. Even community noters cant read. Hey dummies: Do you know what FREE market means"
X Link 2024-05-12T20:14Z [----] followers, [---] engagements

"openai did not leave much room for google team to iterate"
X Link 2024-05-13T22:45Z [----] followers, [---] engagements

"@deliprao @suchenzang The tokens reads straight from all those Simplified Chinese porn/gambling sites hosted in Taiwan Cambodia and Myanmar targeting PRC nationals for loan sharking and kidnapping"
X Link 2024-05-14T00:12Z [----] followers, [---] engagements

"gpt-4o used free adult video to write a song"
X Link 2024-05-14T06:53Z [----] followers, [---] engagements

"introduce 1pondo (US registered JP porn company) Japan 1pondo is japan's most famous scenic region"
X Link 2024-05-14T06:56Z [----] followers, [---] engagements

"@yentinglin56 @OpenAI whats interesting is these tokenization artifacts are almost completely absent in zh-tw or in ja"
X Link 2024-05-14T09:18Z [----] followers, [--] engagements

"@hu_yifei they murder their ads"
X Link 2024-05-14T17:49Z [----] followers, [---] engagements

"@tomolesnevich @firstadopter nvidia is definitely competing vs netflix for talents. who wants to suffer traffic on san thomas to go into office"
X Link 2024-05-16T01:37Z [----] followers, [--] engagements

"@JShpigler @dnystedt @dylan522p @fabknowledge gen6 broadcom gen7 mtk gen8 ky didnt talk about pixel/mobile"
X Link 2024-05-16T04:07Z [----] followers, [--] engagements

"@HamelHusain @Hermes_Paris please sponsor this; we need a bag to hold the agi"
X Link 2024-05-17T21:24Z [----] followers, [---] engagements

"openai deepfaked scarlett johansson's voice. potential apple deal now is in jeopardy"
X Link 2024-05-21T00:04Z [----] followers, [---] engagements

"@ETH_Zurich3773 @NyanCatpital @credenzaclear2 attention"
X Link 2024-05-21T23:30Z [----] followers, [---] engagements

"false dichotomy. toyota cars are both infinitely durable and infinitely easy to repair. team hilux. $5 - Talked to John Ternus - Head of Hardware Engineering at Apple and it was interesting hearing straight from the top why the iPhone is harder to repair. Take a listen https://t.co/O9QsQOx4SP $5 - Talked to John Ternus - Head of Hardware Engineering at Apple and it was interesting hearing straight from the top why the iPhone is harder to repair. Take a listen https://t.co/O9QsQOx4SP"
X Link 2024-05-29T18:31Z [----] followers, [---] engagements

"knowledge graphs attracts data scientists like moths to flames. sounds perfect on paper and most often the only output is a presentation on complexity to impress other data scientists. There are [--] types of Liu 😂 https://t.co/tOTwX7MNxp There are [--] types of Liu 😂 https://t.co/tOTwX7MNxp"
X Link 2024-05-30T19:42Z [----] followers, [----] engagements

"my framework your framework"
X Link 2024-06-01T05:22Z [----] followers, [----] engagements

"finally caught up w/ some of the news. tldr; finance and fundamental research isn't a forte for tech influencers regardless of how many ipo lottery they landed"
X Link 2024-06-01T05:55Z [----] followers, [---] engagements

"value is hard to create. also developers are expensive is a very recent phenomenon The End of Software https://t.co/JWg6QYqLzO The End of Software https://t.co/JWg6QYqLzO"
X Link 2024-06-01T17:46Z [----] followers, [---] engagements

"@Samhanknr tech platforms consolidate as ecosystems mature. and in larger companies theres usually platform teams that helps the consolidation process"
X Link 2024-06-01T21:17Z [----] followers, [--] engagements

"@NyanCatpital salesforce results implied that zirp pacman biz model aint gonna work now and cant find acquisition targets that clears hurdles. the transition from zirp growth to capital structure play is gonna hurt for a while"
X Link 2024-06-02T07:45Z [----] followers, [--] engagements

"Ai engineering anti patterns and code smells: - javascript - camelCase - needle in haystack - app observability vs langsmith/mlflow Whether you're a hiring manager for a SWE moving into AI Engineering a unspoken skill you need to understand is data literacy. I found it hard to figure out what it truely meant so I did it inverted [--] way you'll fuck it up. https://t.co/H2VHrqG462 * Ignoring Data Quality Whether you're a hiring manager for a SWE moving into AI Engineering a unspoken skill you need to understand is data literacy. I found it hard to figure out what it truely meant so I did it"
X Link 2024-06-04T15:14Z [----] followers, [----] engagements

"@abcampbell poorer. east asian countries and city states all industrialized under authoritarian rules. hong kong japan korea singapore taiwan"
X Link 2024-06-04T23:57Z [----] followers, [--] engagements

"@Voyage_AI_ @OpenAI @cohere Do you have sample codes to run and reproduce the benchmarks"
X Link 2024-06-06T00:23Z [----] followers, [--] engagements

"this is a typical misalignment between tech enthusiasts vs the business of software. buyers do not make decisions based on academic benchmarks parameter counts or leader boards. buyers care about SLA security compliance support roadmaps etc. It has to be illegal to say that IBM has a better foundation model offering than OpenAI right @FTC hello https://t.co/xGYteMMDAU It has to be illegal to say that IBM has a better foundation model offering than OpenAI right @FTC hello https://t.co/xGYteMMDAU"
X Link 2024-06-06T06:03Z [----] followers, [---] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing