tweet_id stringlengths 10 19 | full_text stringlengths 16 359 | expanded_url stringlengths 43 52 | embeddings sequencelengths 256 256 |
|---|---|---|---|
1865721226620236176 | Making MLX run on Windows, natively, not WSL or MinGW. https://t.co/qOn9phNf7e | https://twitter.com/i/web/status/1865721226620236176 | [
0.5612192153930664,
-1.3742122650146484,
-1.7414368391036987,
-1.8779406547546387,
0.7156270146369934,
1.4436615705490112,
0.9565522074699402,
-0.37011098861694336,
0.5542045831680298,
0.8383215665817261,
-1.249984622001648,
0.14407220482826233,
-1.1919058561325073,
0.36918655037879944,
... |
1865765748209062099 | @zcbenz Interesting!
It does run, because I usually convert big models on Linux servers.
But there is some digging to do for full support
Great thing is that MLX is built using Cpp so it’s possible. | https://twitter.com/i/web/status/1865765748209062099 | [
0.3261999487876892,
1.2048708200454712,
-1.0372496843338013,
-2.4511358737945557,
1.3667168617248535,
2.3936307430267334,
-0.5775288343429565,
-2.0375192165374756,
-0.2729433476924896,
0.13118717074394226,
-0.4765159785747528,
0.6047537326812744,
0.08176794648170471,
0.33885231614112854,
... |
1865482724351287420 | If you have a Macbook Pro M series with at least 64 GB's of RAM you can now run a GPT-4 level LLM locally!
1. Install @ollama
2. Open your terminal and run ollama pull llama3.3
3. Then ollama run llama3.3 "your prompt"
Your own personal AI is here! https://t.co/jakuVlMteE | https://twitter.com/i/web/status/1865482724351287420 | [
0.6010013222694397,
-0.7031824588775635,
-1.763893723487854,
-1.525599718093872,
1.0397119522094727,
1.0617501735687256,
0.33449459075927734,
-0.06600143015384674,
-0.025183215737342834,
0.19507016241550446,
-1.8157137632369995,
-0.8436480760574341,
-2.0065338611602783,
-1.1084907054901123... |
1865081419015352689 | Gemini-exp-1206, our latest Gemini iteration, (with the full 2M token context and much more) is available right now for free in Google AI Studio and the Gemini API.
I hope you have enjoyed year 1 of the Gemini era as much as I have. We are just getting started : ) | https://twitter.com/i/web/status/1865081419015352689 | [
-0.12989267706871033,
0.4693562686443329,
-2.1581103801727295,
-0.6392908692359924,
0.9318581223487854,
2.75298810005188,
0.34011030197143555,
-0.3327130377292633,
1.3625649213790894,
1.5997809171676636,
-1.8230633735656738,
-1.373061180114746,
-1.2840580940246582,
0.574829638004303,
-0.... |
1865441145788199051 | Pro Tip: if you upgrade from ChatGPT Plus to Pro near the end of your billing cycle you only pay a percentage of $200 for the remaining days.
So you can try unlimited o1 for much less. | https://twitter.com/i/web/status/1865441145788199051 | [
-1.6345093250274658,
3.0760347843170166,
-1.6222363710403442,
-3.0461575984954834,
1.0703048706054688,
1.5225337743759155,
1.4533432722091675,
-0.5846421718597412,
1.2168066501617432,
0.6220940947532654,
-1.2438212633132935,
-1.1582493782043457,
0.18684372305870056,
-1.9802707433700562,
... |
1865632091427520591 | nbsanity now has a bookmarklet https://t.co/VSIrhDXrVS
Thanks to @OAustegard
It's a static server that renders public Jupyter notebooks with Quarto | https://twitter.com/i/web/status/1865632091427520591 | [
1.6696345806121826,
-0.7492926716804504,
-3.3341684341430664,
-2.843214988708496,
0.6583613753318787,
1.5686211585998535,
1.338955283164978,
-0.5099698901176453,
-2.402035713195801,
-0.1417120099067688,
-1.0858708620071411,
-0.5225531458854675,
-0.7961291670799255,
-0.735790491104126,
0.... |
1865534628691677683 | Thanks @_akhaliq! Florence-VL is the first large multimodal model equiped with the popular Florence-v2 vision encoder.
The combination is NOT deliberate, but originated from a deep study based on our proposed visual-semantic alignment metric for different vision encoders and… https://t.co/xSaOgDNfCP https://t.co/0McPg... | https://twitter.com/i/web/status/1865534628691677683 | [
1.3361756801605225,
0.80546635389328,
-3.058751344680786,
-2.517049789428711,
1.0407170057296753,
0.7701636552810669,
0.897321343421936,
1.0834769010543823,
-0.548965334892273,
0.7078219652175903,
-0.8652453422546387,
0.1226915493607521,
-0.7850121259689331,
-0.05312849208712578,
0.79897... |
1865515272032882723 | “we just announced that we are building a 2 gigawatt+ data center in louisiana that we are going to use to train future versions of LLAMA”
that’s the same power consumed by 1.6 million homes, roughly as many as there are in georgia
https://t.co/TZYXMSe12Q | https://twitter.com/i/web/status/1865515272032882723 | [
0.21270334720611572,
0.8135819435119629,
-0.4223920702934265,
-0.45993760228157043,
0.8071768283843994,
1.9780195951461792,
0.6146073937416077,
-1.0449175834655762,
0.23583240807056427,
1.1943066120147705,
0.2898847460746765,
0.06064888462424278,
-1.2217674255371094,
-1.6382408142089844,
... |
1865522970241708391 | https://t.co/gsnXHNxFGA | https://twitter.com/i/web/status/1865522970241708391 | [
5.135742664337158,
-2.3667564392089844,
-3.507371664047241,
-4.431793212890625,
1.0651887655258179,
-1.5854073762893677,
1.7353678941726685,
-0.3281850814819336,
1.751423716545105,
-2.9655418395996094,
0.43927571177482605,
1.7863025665283203,
-0.20935556292533875,
-0.7352243065834045,
0.... |
1865522967041433959 | Instructed Sonnet to generate a nice readme for my perplexity-search repo. It now looks like a proper project even though it's just making an API call https://t.co/cn6Ms593y7 | https://twitter.com/i/web/status/1865522967041433959 | [
-0.9418262243270874,
0.9369025230407715,
-3.8023552894592285,
-1.6584885120391846,
0.39898064732551575,
0.7793837189674377,
-0.3017638623714447,
-0.8334254026412964,
0.37571975588798523,
0.525952160358429,
-2.2959601879119873,
-1.10304856300354,
-1.2160450220108032,
0.10687042027711868,
... |
1865444655342506155 | Running @exolabs AI cluster with @Raspberry_Pi 4.
exo leverages any spare hardware you have available and puts it to use on AI workloads by splitting them across all devices,
You know you've built something magical when you're continually surprised by what people do with it. https://t.co/3Li0Cu6sB7 https://t.co/3xuon... | https://twitter.com/i/web/status/1865444655342506155 | [
-0.2046113759279251,
0.013560152612626553,
-1.6834030151367188,
-2.423550844192505,
1.0437549352645874,
0.8605743646621704,
0.1702089160680771,
0.17842787504196167,
-0.2965546250343323,
0.7136004567146301,
-0.8219081163406372,
-0.7049912810325623,
-1.514407753944397,
0.09238770604133606,
... |
1865589927540342801 | https://t.co/u6pljatcvL | https://twitter.com/i/web/status/1865589927540342801 | [
1.916953206062317,
0.309400349855423,
-2.177018404006958,
-3.3140716552734375,
-1.403745412826538,
1.069481372833252,
0.21788063645362854,
2.750255823135376,
-2.210149049758911,
0.313692569732666,
-1.3859862089157104,
-2.5216517448425293,
-3.0636069774627686,
-1.3063502311706543,
-1.1236... |
1865589925170528646 | "Take control of your AI agents" https://t.co/ohYy7ymC7l | https://twitter.com/i/web/status/1865589925170528646 | [
-0.9897646903991699,
-1.0810028314590454,
-2.733896017074585,
-0.9037990570068359,
0.29021933674812317,
-0.38527727127075195,
-1.931527018547058,
1.922024130821228,
1.0635361671447754,
0.7301600575447083,
-1.3413044214248657,
-0.2074512243270874,
-2.0266358852386475,
-1.1959940195083618,
... |
1865305243048718519 | Sinterklaas kwam een dag te laat langs. Op zijn mijter stond in gouden letters ‘De Standaard” geborduurd. Op zijn staf kon je ook nog in kronkelende kapitalen ‘De Letteren’ lezen. De goede man haalde onderstaand artikel uit zijn zak.
https://t.co/2MUIlqKmoy | https://twitter.com/i/web/status/1865305243048718519 | [
0.6869345307350159,
-3.7449984550476074,
-4.150768280029297,
-3.087956666946411,
1.9076157808303833,
0.05204140767455101,
2.476619243621826,
3.5545766353607178,
-1.893348217010498,
0.20842379331588745,
0.6738370060920715,
0.07491686940193176,
-0.4045279920101166,
0.81241774559021,
1.1795... |
1865419965190328488 | currently having my mind blown reading this book in a coffee shop. seriously wtf, wish i found this sooner. https://t.co/ZAfmw820d9 | https://twitter.com/i/web/status/1865419965190328488 | [
-3.5176889896392822,
-1.0426421165466309,
-0.9501897692680359,
-0.7234265804290771,
0.8320796489715576,
3.4448158740997314,
1.3029401302337646,
-0.620251476764679,
1.4819210767745972,
1.9641950130462646,
0.3239336609840393,
-0.1714840680360794,
-0.774284839630127,
0.46205654740333557,
0.... |
1865010068204486915 | Florence-VL challenges the status quo in Vision-Language Models!
Even Google's PaliGemma 2 dropped yesterday (still using SigLIP), Florence-VL shows us there might be a better way! Here's why this matters 👇
📸 Look at this revealing visualization:
Left: Three test images… https://t.co/IkBTsZSNBW | https://twitter.com/i/web/status/1865010068204486915 | [
-0.6268230676651001,
-0.16394264996051788,
-2.2458713054656982,
-1.3855488300323486,
0.440878689289093,
1.2784671783447266,
0.5558773279190063,
-0.8045994639396667,
0.04955052211880684,
0.07992351800203323,
-0.183809295296669,
-0.6618932485580444,
-0.5566936135292053,
-0.5757043957710266,
... |
1865399992690544866 | @willmcgugan You can hit https://t.co/6rsWJpgkNp, then grab response["info"]["version"] | https://twitter.com/i/web/status/1865399992690544866 | [
0.43823763728141785,
0.9097719788551331,
-2.019831657409668,
-2.493041515350342,
2.854469060897827,
0.6061225533485413,
-0.2034129649400711,
-0.09327024966478348,
-0.2897675037384033,
-1.0327593088150024,
-1.6945724487304688,
0.7009143829345703,
-1.4448578357696533,
0.37474170327186584,
... |
1865188121060712479 | Command:
mlx_lm.generate --model mlx-community/Llama-3.3-70B-Instruct-4bit --max-tokens 256 --prompt "Do avocado trees grow in the bay area?"
Thanks to @Prince_Canuma for converting the models. A bunch of different quants (3, 4, 6, 8, etc) are up the MLX Community:… | https://twitter.com/i/web/status/1865188121060712479 | [
1.34731924533844,
0.24823150038719177,
-1.8499587774276733,
-2.5581798553466797,
2.191244125366211,
1.7003554105758667,
-0.4737775921821594,
-1.249107837677002,
-0.38907381892204285,
0.7452213168144226,
-0.90984708070755,
-0.05473489686846733,
-0.6675773859024048,
-0.3025948405265808,
0.... |
1865199571988611382 | 🐍📺 Using Data Classes in Python [Video] — https://t.co/cVTmP8wjDR
#python https://t.co/pm7AaPnLjY | https://twitter.com/i/web/status/1865199571988611382 | [
1.8302737474441528,
0.5730268359184265,
-2.9225635528564453,
-2.7021970748901367,
0.016618266701698303,
0.8092389702796936,
0.5643011927604675,
0.8925224542617798,
-1.6730509996414185,
-1.435531735420227,
-2.0370800495147705,
0.8970026969909668,
-1.6527938842773438,
1.7432680130004883,
-... |
1688849233967730688 | Some research informed thoughts on adding brief sprints to low-intensity (LIT) sessions:
1. Keep the duration down to 3 to 4 seconds! This is the typical "Alactic" duration. Even extending to 6-8 seconds will result in significant lactate production. If you drop in a block of… | https://twitter.com/i/web/status/1688849233967730688 | [
-1.973872423171997,
2.130939483642578,
-0.21569158136844635,
-2.911151170730591,
0.1612633317708969,
1.0122284889221191,
-0.08207631856203079,
-2.3450734615325928,
-0.7517605423927307,
1.0894562005996704,
0.21033510565757751,
0.2984728515148163,
-2.512159585952759,
-0.05329269915819168,
... |
1865170551368519958 | omg that slide took me probably a whole evening lol https://t.co/aTuIFrTD7a | https://twitter.com/i/web/status/1865170551368519958 | [
-1.129042625427246,
-1.7932120561599731,
-2.5504777431488037,
-1.6982027292251587,
-0.1557493954896927,
0.7028782367706299,
1.5976227521896362,
0.3295997977256775,
-0.13192510604858398,
1.5060228109359741,
-0.6295570135116577,
-0.8269705176353455,
-1.814416766166687,
-0.6586761474609375,
... |
1865203069639709079 | "Mastering Applied AI, One Concept at a Time" https://t.co/ByXiy0rJzN | https://twitter.com/i/web/status/1865203069639709079 | [
-0.451631098985672,
-0.30567657947540283,
-3.767235040664673,
-3.0008809566497803,
0.9573404788970947,
-0.4487035572528839,
-0.8095642328262329,
0.16475635766983032,
-0.622283399105072,
-0.0749366283416748,
-0.2562319338321686,
0.38589489459991455,
-2.4642322063446045,
0.7463423609733582,
... |
1865203078040826254 | https://t.co/71G609LHhe | https://twitter.com/i/web/status/1865203078040826254 | [
-1.570193886756897,
-1.022013783454895,
-2.4261319637298584,
-1.5650354623794556,
0.8692456483840942,
1.771499752998352,
3.304534912109375,
0.5371256470680237,
0.7225611805915833,
-0.6320534348487854,
-0.6217443943023682,
-1.4915125370025635,
-0.41789355874061584,
-3.0929741859436035,
0.... |
1865034787582251340 | PaliGemma2 for image to JSON data extraction
- used google/paligemma2-3b-pt-336 checkpoint; I tried to make it happen with 224, but 336 performed a lot better
- trained on A100 with 40GB VRAM
- trained with LoRA
colab with complete fine-tuning code: https://t.co/M1lbYXQUg6 https://t.co/DHNHGePaqM | https://twitter.com/i/web/status/1865034787582251340 | [
0.34394383430480957,
0.7778353095054626,
-2.895763635635376,
-2.5097544193267822,
1.137946367263794,
1.3457980155944824,
0.8381218910217285,
0.3442422151565552,
0.8346719741821289,
-0.6113961338996887,
-0.7050164341926575,
-0.510120689868927,
-1.2011185884475708,
-1.0977107286453247,
0.2... |
1865187697146700273 | Llama 3.3 70B 4-bit runs nicely on a 64GB M3 Max with in MLX LM (~10 toks/sec). Would be even faster on an M4 Max.
Yesterday's server-only 405B is today's laptop 70B: https://t.co/ssxITH5ggT https://t.co/vgEn2E5WS8 | https://twitter.com/i/web/status/1865187697146700273 | [
0.9599223732948303,
0.242132768034935,
-1.5308445692062378,
-2.3684821128845215,
2.433318853378296,
2.118936538696289,
1.3987523317337036,
-1.2276803255081177,
1.3235549926757812,
-0.07478565722703934,
-1.1558858156204224,
0.1721528172492981,
-1.4624801874160767,
-0.9021667838096619,
1.0... |
1865198113188643197 | https://t.co/zGpeO2mJLi | https://twitter.com/i/web/status/1865198113188643197 | [
3.0623300075531006,
-0.7782238125801086,
-4.676839351654053,
-1.9230809211730957,
0.6540589928627014,
0.11029358208179474,
1.5960545539855957,
0.923703134059906,
0.5802040696144104,
0.2868264317512512,
0.29931458830833435,
-0.7229024171829224,
-2.429213762283325,
0.2884608507156372,
1.09... |
1865162836948791299 | Quick community showcase!
@aditshah00 added an example of running marimo inside @modal_labs, bringing serverless cloud computing power to interactive notebooks! 💪 https://t.co/SDUycwoTOE | https://twitter.com/i/web/status/1865162836948791299 | [
1.9286004304885864,
0.43670013546943665,
-2.8277153968811035,
-3.910433769226074,
0.7495779991149902,
2.1466991901397705,
1.0774458646774292,
0.1068810373544693,
-0.8883339166641235,
2.1014482975006104,
-1.1670249700546265,
0.2716890573501587,
-0.16907916963100433,
-0.21337918937206268,
... |
1865184209754525943 | One of the best use cases for geospatial data is examining environmental factors.
Here's a list of my favourite geospatial environmental datasets: https://t.co/HhN7GuqHG9 | https://twitter.com/i/web/status/1865184209754525943 | [
4.145713806152344,
0.8961296677589417,
-2.2549750804901123,
-2.5679943561553955,
-0.8873301148414612,
-0.15280723571777344,
3.080845832824707,
-1.1227411031723022,
0.49494320154190063,
0.47064095735549927,
1.5917214155197144,
0.8096921443939209,
-1.2325468063354492,
-1.7758088111877441,
... |
1865203415342612732 | moka-py
A high performance caching library for Python written in Rust.
https://t.co/eGICfwfksO | https://twitter.com/i/web/status/1865203415342612732 | [
2.156484365463257,
-0.2064889818429947,
-1.7748850584030151,
-2.762277603149414,
0.9191105961799622,
2.2378172874450684,
-0.4639986753463745,
-1.1286370754241943,
-0.5283827185630798,
0.6821969151496887,
-0.2640315592288971,
1.4122806787490845,
0.18080246448516846,
1.5311377048492432,
0.... |
1865073523896516950 | https://t.co/fomVkelxbx | https://twitter.com/i/web/status/1865073523896516950 | [
0.5382225513458252,
-0.8595911860466003,
-3.2280924320220947,
-4.150956630706787,
-0.1429397612810135,
0.2698945105075836,
1.5819551944732666,
2.1130638122558594,
0.9035160541534424,
-0.2464761734008789,
-1.0476776361465454,
0.7037813067436218,
-1.0358021259307861,
-0.22533230483531952,
... |
1865073353196794156 | I have got to praise Alibaba for mPLUG DocOwl.
What an amazing state of the art tool, and they have open sourced their WHOLE pipeline. Code, Dataset, Weights, Everything.
Bravo! Links in thread. https://t.co/zidwL2ZlGx | https://twitter.com/i/web/status/1865073353196794156 | [
0.44533610343933105,
-0.4145002067089081,
-0.9867087602615356,
-2.637251138687134,
1.4870021343231201,
2.686047077178955,
0.3960570693016052,
0.5683489441871643,
-1.250717043876648,
-0.09192276000976562,
-1.3320353031158447,
-0.01903865486383438,
-0.10840732604265213,
-0.8812541961669922,
... |
1865184608070824116 | Data flywheels are the secret sauce for AI-driven products.
Use your users' interactions to continuously improve. The more users, the better the product, the more users.
Netflix and Spotify nailed it. You should too.
https://t.co/wClR6kBYCW | https://twitter.com/i/web/status/1865184608070824116 | [
-2.240278482437134,
0.5952473282814026,
-1.1407394409179688,
-2.8183372020721436,
0.7800147533416748,
1.5294750928878784,
-1.0255076885223389,
-0.0164325013756752,
0.7742634415626526,
1.2237919569015503,
-0.010188041254878044,
-1.2985681295394897,
-1.372240662574768,
0.48738428950309753,
... |
1865096566467686909 | today we are announcing reinforcement finetuning, which makes it really easy to create expert models in specific domains with very little training data.
livestream going now: https://t.co/ABHFV8NiKc
alpha program starting now, launching publicly in q1 | https://twitter.com/i/web/status/1865096566467686909 | [
0.06157713755965233,
2.413179397583008,
-1.3406474590301514,
-3.0811715126037598,
0.8652238249778748,
1.6039941310882568,
-0.5466339588165283,
1.1087727546691895,
0.08753564208745956,
0.24537315964698792,
-1.8313907384872437,
0.05801395699381828,
-0.5019732117652893,
0.8649920225143433,
... |
1865123445220028910 | Coming to MLX 🚀 https://t.co/YQhS0IDqtr https://t.co/1ri9OenvdZ | https://twitter.com/i/web/status/1865123445220028910 | [
1.1849926710128784,
-1.2456402778625488,
-3.257899045944214,
-2.4553170204162598,
0.041203975677490234,
0.3712877035140991,
2.2239174842834473,
1.8722413778305054,
0.09239891916513443,
-0.4299595355987549,
-0.3285350203514099,
0.3325439393520355,
-2.206737518310547,
0.30403247475624084,
... |
1865088634342523169 | "a new 70B model that delivers the performance of our 405B model" is exciting because I might just be able to run a quantized version of the 70B on my 64GB Mac - looking forward to some GGUFs of this https://t.co/6X9wDesnEG | https://twitter.com/i/web/status/1865088634342523169 | [
0.16447895765304565,
0.6904184818267822,
-0.7005704045295715,
-2.887962579727173,
2.117222785949707,
2.8938915729522705,
-0.21498848497867584,
-1.1959648132324219,
1.2715234756469727,
-0.30672013759613037,
-0.12930341064929962,
-0.6708284616470337,
-1.0504026412963867,
-0.39967289566993713... |
1865070330286403846 | Belg is financieel gelukkig als hij meer dan 5.500 euro netto per maand verdient
https://t.co/bcg0AdyMs3 | https://twitter.com/i/web/status/1865070330286403846 | [
2.876303195953369,
-2.0476644039154053,
-2.529730796813965,
-3.917426586151123,
2.944241523742676,
1.445269227027893,
0.44565364718437195,
1.330523133277893,
-0.7241898775100708,
0.7630322575569153,
1.0895658731460571,
0.8396067023277283,
0.149701327085495,
0.2999117374420166,
1.31360220... |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 9