Manu Romero @mrm8488 Twitter profile

Pinned Tweet

Manu Romero

@mrm8488

8 years

Nuestra joya P. R. E.

5

4

57

Last Seen Profiles

@bokeplokalmalam

@paul_brickley

@LisaASchulz2

@luckywithlola

@18lewis67

@dashtiabataa

@bokeplokalmalam

@victorsavkin

@Jogigios

@ashley_har45753

@_Inari_Kohaku_

@EndMassMigrate

@Jason_Stollman

@ozzu_dq

@hs_visio

@EyadQunaibi1

@yahyaahmed91151

@WaterwayProject

@vodkahist

@OpusArtsEvents

@MichalAsa189

@bokeplokalmalam

@lisaairplanes

@gadoro_mattero

@defeatedoo

@Alexruoff

@IchabodVB

@bokeplokalmalam

@Khaled_ALmuzani

@mytraces17

@gordperks

@slainbyelf

@bokeplokalmalam

@MichalMoch87

@NaughtyCinnabun

Manu Romero

@mrm8488

2 years

I fine-tuned @BigscienceW 🌸 BLOOM (7B) on Standford's #Alpaca 🦙 dataset using @huggingface PEFT (LoRA), and the results are awesome! Any name suggestion for the model?

52

147

901

Manu Romero

@mrm8488

2 years

Celebrating 🎉 300 open source AI models at the @huggingface model hub!!! 🎉🤗

15

26

535

Manu Romero

@mrm8488

4 years

📒Colab for: Fine-tuning a deep-learning model (bert-base-cased) with @huggingface 🤗Datasets library and @PyTorch (included in the quick start guide) on glue / mrpc

example_fine-tuning-hf_datasets.ipynb

Colaboratory notebook

colab.research.google.com

2

89

395

Manu Romero

@mrm8488

1 year

I will release a Colab Notebook to fine-tune Falcon 40B on a single A100 (40Gb) GPU using #QLoRA .

13

30

343

Manu Romero

@mrm8488

4 years

Data augmentation in #NLP is a very insteresting research field. I created this colab where @tobias_sterbak shows Data augmentation with ( @huggingface ) transformer models for #NER

NER_data_augmentation.ipynb

Colaboratory notebook

colab.research.google.com

4

75

325

Manu Romero

@mrm8488

6 months

Minimal Fine-tuning example on a Colab (Pro) Notebook (A100 - 40GB) for @AI21Labs #Jamba

fine-tune-Jamba.ipynb

Colaboratory notebook

colab.research.google.com

6

47

309

Manu Romero

@mrm8488

8 years

@iunida no os dais cuenta de que estáis haciendo el ridículo felicitando a un dictador en pleno S. XXI??

12

55

246

Manu Romero

@mrm8488

10 months

It seems 🤖Reinforcement learning👾 is back! Very good resources for getting started on it from @huggingface and @ThomasSimonini

Welcome to the 🤗 Deep Reinforcement Learning Course - Hugging Face Deep RL Course

huggingface.co

0

45

252

Manu Romero

@mrm8488

2 years

Many of you asked me for a Colab Notebook 📒 to train a model to translate from SQL 💽 to Natural Language 🗣️ (and vice versa). Tomorrow I will release it using Google's T5 and @huggingface ecosystem. #nlp #nlg #seq2seq #tuto

7

29

248

Manu Romero

@mrm8488

1 year

🧵When using #LoRA it is important to apply it to ALL `Linear` layers of the model to get similar results to "full fine-tuning." If you are using @huggingface /peft library, you will see some LoRA configs like the following

6

37

224

Manu Romero

@mrm8488

1 year

I fine-tuned #Falcon 🦅7B () on a Spanish instruction dataset using #QLoRA (4bit), and the results are promising! Thanks to @younesbelkada for the support!

13

26

216

Manu Romero

@mrm8488

2 years

Repo: Alpaca + BLOOM = Alpacoom

Manu Romero

@mrm8488

2 years

I fine-tuned @BigscienceW 🌸 BLOOM (7B) on Standford's #Alpaca 🦙 dataset using @huggingface PEFT (LoRA), and the results are awesome! Any name suggestion for the model?

52

147

901

8

46

210

Manu Romero

@mrm8488

5 months

Colab 📒 Notebook to fine-tune 💅🏽 @GoogleAI #PaliGemma vision-language model 👓🔠🧠 on a free T4 VM!

finetune_paligemma.ipynb

Run, share, and edit Python notebooks

colab.research.google.com

2

54

211

Manu Romero

@mrm8488

4 years

T5 fine-tuned on #wikiSQL as translation task. I am impressed with the results! Ofc, the model will be availiable at @huggingface #NLP #SQL .

10

37

196

Manu Romero

@mrm8488

1 year

🚨 New Model Alert 🚨 After the success of FalCoder 🦅👩‍💻, I have created #LLama2Coder 🦙👩‍💻. A Llama-2 (7B) model fine-tuned on the #CodeAlpaca dataset 📚 for #codegen . Download it from the 🤗 @huggingface hub:

mrm8488/llama-2-coder-7b · Hugging Face

huggingface.co

4

49

173

Manu Romero

@mrm8488

4 years

I turned this great post by @txustice about semantic search and @huggingface Transformers into a Colab:

Sematinc_Search_Spanish.ipynb

Colaboratory notebook

colab.research.google.com

2

31

167

Manu Romero

@mrm8488

2 years

In my next Colab Notebook tutorial 📙👩‍🏫 you will learn to fine-tune Google's T5 (small) 🧠to translate from Natural Language 🗣️ to SQL 💾 and vice versa (multitask) ↔️using @huggingface 🤗! #nlp #sql #text2text #tuto

4

20

154

Manu Romero

@mrm8488

2 years

In case you were asking about create a #docker container via docker-compose with #GPU capabilities for your #ML app.

4

26

147

Manu Romero

@mrm8488

4 years

So, AFAIK there are many sentiment analysis models bust vast majority of them are binary (negative, positive). So following the great tutorial of @omarsar0 (based on @marrrcin one) I tweaked a bit the model (6 emotions) & uploaded it to @huggingface #NLP

4

37

148

Manu Romero

@mrm8488

3 years

@julien_c `nvidia-smi` showing this:

8

5

149

Manu Romero

@mrm8488

10 months

🚨Thrilled to announce our new model🚨 🐍Mamba🐍 (2.8B) fine-tuned on OpenHermes dataset for instruction following! Check it out in the @huggingface model hub:

clibrain/mamba-2.8b-instruct-openhermes · Hugging Face

huggingface.co

Manu Romero

@mrm8488

10 months

At @ClibrainAI we fine-tuned Mamba (2.8B) 🐍on @huggingface H4️⃣ no robots (⛔🤖) for chat / instruction following! Enjoy! 🥂

2

8

45

7

26

141

Manu Romero

@mrm8488

1 year

🚨Thrilled to announce FalCoder-7B🚨: #Falcon 🦅 fine-tuned on CodeAlpaca20k Dataset 📚 for #code instruction completion 👩‍💻 Some examples below 👇

mrm8488/falcoder-7b · Hugging Face

huggingface.co

Manu Romero

@mrm8488

1 year

Falling today... #Falcon #LLM

1

2

21

7

27

141

Manu Romero

@mrm8488

11 months

🚨 New model alert 🚨 I fine-tuned @MistralAI 7B on @huggingface H4⃣ "No robots" datasets for instruction following downstream task. Here are the model card and weights: and an example of the model talking about the weekend topic ( @sama )

7

19

136

Manu Romero

@mrm8488

2 years

With the latest @GoogleColab pricing updates, having a PRO Plus subscription, if I set up a VM with an A100 GPU, I "only" can use it for 28.47h... 😯🙀🫢 Before that, you got an A100 rarely, but you were guaranteed a V100 GPU for a whole month!

2

8

130

Manu Romero

@mrm8488

4 years

Maybe the smallest model for question generation (English) in the @huggingface model hub: shared BERT2BERT(small) fine-tuned on SQUAD for #QG . ❔✍️ Model size: 127 MB Some examples: 👇

1

18

134

Manu Romero

@mrm8488

7 months

I found this hidden gem 💎 in the format of colab. An end-to-end example by @huggingface H4 team to create a synthetic dataset of web layouts to html/css code! 📔 Colab: 📚 Dataset: 📰 Post:

Create synthetic web screenshots and their associated HTML code with Mistral and Deepseek Code...

Colaboratory notebook

colab.research.google.com

1

33

132

Manu Romero

@mrm8488

4 years

And now, I inverted the process: From #SQL to Natural language (English). You can find the model at @huggingface #NLP #Text2Text #T5

Manu Romero

@mrm8488

4 years

T5 fine-tuned on #wikiSQL as translation task. I am impressed with the results! Ofc, the model will be availiable at @huggingface #NLP #SQL .

10

37

196

3

35

133

Manu Romero

@mrm8488

2 years

Glad to see that my "noise" can help other NLP/G researchers 🤗

1

10

125

Manu Romero

@mrm8488

7 years

@pnique @GirautaOficial @CiudadanosCs Y el zasca llegó...

6

32

110

Manu Romero

@mrm8488

11 months

Underrated resource by @AiEleuther

1

22

123

Manu Romero

@mrm8488

3 years

I will finish the year with 250+ language models 🗣️🧠 on the @huggingface model hub! 🎉🥂. Thanks to @NarrativaAI for the support (specially @alberto_at_nrt ) and @julien_c from HF.

2

13

120

Manu Romero

@mrm8488

4 years

📒Colab for NER: end-to-end example with ( @huggingface )🤗Datasets and 🤗Transformers scripts. #NLP #NER

fine-tuning on UDPOS English for NER.ipynb

Colaboratory notebook

colab.research.google.com

1

33

119

Manu Romero

@mrm8488

11 months

🧠🍋 LIMSTRAL: #Mistral fine-tuned on the #LIMA dataset for instruction following downstream task. MTBench eval is WIP

mrm8488/limstral-7B-v0.1 · Hugging Face

huggingface.co

3

17

119

Manu Romero

@mrm8488

7 months

🚨 Important 🚨 For those using LoRA/QLoRA for fine-tuning their model: If you add special tokens to the tokenizer, such as <special_token1>, <special_token2>, etc., ensure you also add the embed_tokens module to your LoRA config target modules. Without this, the embeddings for

3

16

119

Manu Romero

@mrm8488

4 years

Low resources languages also deserve its LMs. So proud of releasing #RoBasquERTa : a #RoBERTa like model trained on OSCAR #Basque corpus. You can find it a @huggingface model hub: #AI #ML #DL #NLP

3

35

113

Manu Romero

@mrm8488

2 years

1) Get an LLM (>3B params) 2) Fine-tune it on Instructions dataset/s, CoT 3) Refine it with RLHF 4) Let's compete against #ChatGPT

7

13

113

Manu Romero

@mrm8488

10 months

🚨New #LLM Release🚨 Today, at @ClibrainAI , we release our latest 🐍 Mamba (2.8B) fine-tuning 🐍 and we do it on Sythina-v1.3 dataset: Model ( @huggingface hub): Collection: Enjoy 🍾

🐍 Mamba fine-tuned models - a clibrain Collection

huggingface.co

1

20

112

Manu Romero

@mrm8488

7 years

@MovistarSeries Ponedlo en base hexadecimal y ahorráis un montón de tuits.

1

5

104

Manu Romero

@mrm8488

5 years

Maybe the smallest BERT like model (17 MB) fine tuned on SQuaD 2. cc: @huggingface

3

24

97

Manu Romero

@mrm8488

5 years

My #NLP favourite resources of March: - @Thom_Wolf : An Introduction to Transfer Learning and @huggingface : - @seb_ruder : I discovered his awesome blog - @joeddav : REALM: Knowledge and Transformers () 1/2

Thomas Wolf (HuggingFace): An Introduction to Transfer Learning and...

NLP Zurichhttps://www.meetup.com/NLP-ZurichThomas Wolf: An Introduction to Transfer Learning and HuggingFaceIn this talk I'll start by introducing the recent...

www.youtube.com

3

23

98

Manu Romero

@mrm8488

4 years

If you like #History , #Philosophy and #AI ( #NLP ) I have fine tuned distil #GPT -2 on Meditations - Marco Aurelio. You can find the model on @huggingface hub and ask about the meaning of life (for example)

4

15

98

Manu Romero

@mrm8488

4 years

With the last release of @huggingface it is so easy to log your model metrics while training on @weights_biases and get meaningful graphics

3

14

95

Manu Romero

@mrm8488

3 years

Yup, #Spanish GPT-2 is working!

6

17

95

Manu Romero

@mrm8488

2 years

After +2 years collaborating with @huggingface and almost 300 models in the hub I finally became part of 🤗 Fellows Program!

10

6

92

Manu Romero

@mrm8488

1 year

A new #Falcon tuned model is falling this week...

5

6

93

Manu Romero

@mrm8488

3 years

So at @NarrativaAI we have fine-tuned the first #ByT5 model (small) on the @huggingface model hub. The dataset is: tweets hate speech detection and we got an accuracy of: 97.8 %! Try it out: #NLP #NLP #Text2Text

mrm8488/byt5-small-tweet-hate-detection · Hugging Face

huggingface.co

Hugging Face

@huggingface

3 years

🚀 And merged to Transformers! We are excited to welcome ByT5 as the first tokenizer-free model! 👉All available checkpoints can be accessed on the 🤗hub here: 👇 Demo (on master):

7

82

387

5

15

91

Manu Romero

@mrm8488

1 year

Must read about "Optimizing #LLMs in Production" by my friend Patrick von Platen in the @huggingface blog.

Optimizing your LLM in production

huggingface.co

0

25

90

Manu Romero

@mrm8488

9 months

🚨New Model Alert🚨 Thrilled to announce **Phi-2-coder**. MS 🪟 Phi-2 fine-tuned on codeAlpaca 🦙⌨️ for code 👩‍💻 instruction following Test the model on a free colab: I will also share the weights for running it with #MLX 🍏

mrm8488/phi-2-coder · Hugging Face

huggingface.co

2

16

88

Manu Romero

@mrm8488

2 years

Celebrating 🎉 300 open source AI models at the @huggingface model hub!!! 🎉🤗

2

7

88

Manu Romero

@mrm8488

4 years

4 models to reach 200 at @huggingface model hub. So proud! #NLP #AI

2

10

87

Manu Romero

@mrm8488

2 years

🚨 New multilingual🌏 model 🧠alert🚨 XLM-V is already in the @huggingface 🤗Hub!

facebook/xlm-v-base · Hugging Face

huggingface.co

4

12

83

Manu Romero

@mrm8488

3 years

At @NarrativaAI research we continue working on Personal Information Identification ( #PII ) in #medical / #clinical context ( #Spanish )! As you can see in the image below our #model seems to work fine! We will share it soon via @huggingface model hub! #AI #NLP #research #phrama

0

18

82

Manu Romero

@mrm8488

2 years

Amazon's multilingual review ds is a large dataset of customer reviews in several languages. I reframed it as a binary classification problem and by using #SetFit w/ ONLY 16 examples per class I reached 70% in the test set!

2

15

83

Manu Romero

@mrm8488

4 months

🚀 Just out: Sentence-Transformers 3 is transforming the game! Kudos to @tomaarsen for the stellar update. 🌟 🔥 NEW FEATURE: Train your own Matryoshka embedding models! Want to dive in? I've set up a Colab notebook to get you started right away. Check it out and start creating

0

12

83

Manu Romero

@mrm8488

2 years

@adampickard @_akhaliq

3d-stable-diffusion.ipynb

Colaboratory notebook

colab.research.google.com

2

15

82

Manu Romero

@mrm8488

2 years

🚨 New model alert!🚨 Thrilled to announce a new Spanish Language Model for the ⚖️Legal ⚖️domain! I pre-trained a #BART model (encoder-decoder) on the "Collection of corpora of Spanish legal domain"

mrm8488/bart-legal-base-es · Hugging Face

huggingface.co

7

16

78

Manu Romero

@mrm8488

9 months

Running Mistral-7B fine-tuned on instructions v2 (4-bit quantized) on my Mac thanks to #MLX ⚡️

6

7

76

Manu Romero

@mrm8488

2 years

And finally... 🥁🥁🥁 I reached3⃣0⃣0⃣ open models on the @huggingface 🤗 model hub! 🥳

3

9

77

Manu Romero

@mrm8488

4 years

Model in action!! #Algebra #NLP #T5

Manu Romero

@mrm8488

4 years

Google's T5 solving linear algebra basic equations! #T5 #NLP #Maths #Text2Text #Seq2Seq

2

1

6

0

21

75

Manu Romero

@mrm8488

6 months

I created a minimal Colab Notebook 📙 to Fine-tune LlaVa-1.5-7B 👁️🔤 on the `llava-instruct` dataset!

fine_tune_VLM_LlaVa.ipynb

Run, share, and edit Python notebooks

colab.research.google.com

merve

@mervenoyann

6 months

Ever wanted to learn about fantastic vision language models and how to find and fine-tune them? 🧙🏻 We've just added support to train VLMs like LLaVa in TRL and wrote a walkthrough on vision language models! 🎉 Read about VLMs and SFTTrainer for vision

12

33

190

2

22

75

Manu Romero

@mrm8488

2 years

Distill BLOOM is happening @BigscienceW

bigscience/distill-bloom-1b3 at main

huggingface.co

1

15

73

Manu Romero

@mrm8488

6 months

Thank you so much for the swag @huggingface ( @mervenoyann )

8

5

75

Manu Romero

@mrm8488

1 year

I am very excited to share with you our new model! At @ClibrainAI we believe in #multimodal as the shorter way to reach #AGI , and based on it we have fine-tuned the new @huggingface IDEFICS model (Flamingo-like) to generate MJ/SD-like captions!

6

22

72

Manu Romero

@mrm8488

7 years

@pnique Espero que al traductor no le pagues también en B.

3

6

63

Manu Romero

@mrm8488

3 years

Thrilled to announce the first Spanish #LongFormer model! 🤖 📝📝📝➡️ 📄 It supports sequences of length up to 𝟜𝟘𝟡𝟞! Thanks to BERTIN team, @NarrativaAI and @i_beltagy for their work and support! #nlp

mrm8488/longformer-base-4096-spanish · Hugging Face

huggingface.co

2

21

72

Manu Romero

@mrm8488

3 years

Open to everybody the demo of the most powerful Spanish GPT-2 in the @huggingface model hub!

Spanish GPT-2 - a Hugging Face Space by flax-community

huggingface.co

3

21

71

Manu Romero

@mrm8488

4 years

Your wish is my command! Today I am releasing on @huggingface the first Electra model trained on Spanish corpus 🇪🇸🗣️: ⚡️electricidad-small-discriminator⚡️ #NLP #AI #Spanish

clem 🤗

@ClementDelangue

4 years

@mrm8488 @huggingface Electracidad!

1

2

8

4

18

71

Manu Romero

@mrm8488

9 months

I am starting a series of fine-tuning experiments w/ MS Phi-2. I've begun w/ the HQ instruction/chat dataset "no_robots" (by @huggingface H4)

mrm8488/phi2-ft-no_robots-adapter · Hugging Face

huggingface.co

5

7

67

Manu Romero

@mrm8488

3 years

Thrilled to announce 𝘽𝙄𝙊𝙈𝙀𝘿𝙩𝙧𝙖! the first #electra ⚡️ model pre-trained on CoWeSe 🦠🏥 (Spanish #Biomedical Crawled Corpus) to help in downstream tasks for this domain/s.

mrm8488/biomedtra-small-es · Hugging Face

huggingface.co

1

14

68

Manu Romero

@mrm8488

8 months

Miqu-70B 4bit quant (aka Mistral-medium?) generating on a Colab A100 40GB VM

2

9

66

Manu Romero

@mrm8488

2 years

Dealing with data that has few to no labels? 🤔 Did you know @huggingface #SetFit ? It is an efficient framework for few-shot fine-tuning of Sentence Transformers that achieves high accuracy with little labeled data (8 examples per class). 👇

1

2

66

Manu Romero

@mrm8488

3 years

And the most USEFUL Language Model of 2021 was... 🥁🥁🥁🥁🥁🥁 Google's T5 by @colinraffel etc al.

4

2

66

Manu Romero

@mrm8488

4 years

So now you can play with different fine-pruned models on SQUAD v1, multilingual QA (tydiQA) and SQAD v2. All multilingual models are under 50 MB (Encoder sparsed and quantized) and never under F1=70. Thanks to @SanhEstPasMoi @Thom_Wolf and A.M Rush @huggingface

1

16

67

Manu Romero

@mrm8488

2 years

I trained FLAN-T5 (small) on it

mrm8488/flan-t5-small-finetuned-openai-summarize_from_feedback · Hugging Face

huggingface.co

Tristan Thrush

@TristanThrush

2 years

A new @OpenAI human-feedback dataset is on the Hugging Face Hub! This one is from the "Learning to Summarize from Human Feedback" paper, where the amazing authors trained an RLHF reward model for summarization.

1

40

214

5

10

66

Manu Romero

@mrm8488

2 years

Diffusers 🧨 from @huggingface has been a successful library to work w/ Diffusion models but keep an eye 👀 on the new one called #PEFT 🐰: It allows us to fine-tune Large Language Models efficiently using more modest setups!

GitHub - huggingface/peft: 🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning. - huggingface/peft

github.com

0

8

65

Manu Romero

@mrm8488

4 years

Roberta2Roberta, another #summarization model at @huggingface model hub: An encoder-decoder model initialized on the 'roberta-large' ckpts for both the encoder and decoder and fine-tuned on extreme summarization on the BBC XSum dataset. #NLP

google/roberta2roberta_L-24_bbc · Hugging Face

huggingface.co

0

15

66

Manu Romero

@mrm8488

2 years

Maybe my last model 🧠 before Xmas 🎄. I fine-tuned @BigscienceW BLOOM 🌸 (560M Ckpt) on unnatural instructions 👩‍🏫 dataset for 6k steps and results are better than I expected.

2

7

64

Manu Romero

@mrm8488

4 years

Happy to finish this strange and hard year among top @huggingface contributors!

4

64

Manu Romero

@mrm8488

2 years

Still in the draft phase (metrics must be computed on the test dataset). But, for the most impatient, here you have a #colab 📔 notebook for fine-tuning 🏋️‍♀️ #BLOOM 🌸 on CNN/Dailymail 🗞️ dataset for #summarization 📝

3

10

63

Manu Romero

@mrm8488

1 year

My next talk is a about the IT LLMs Zoo

3

12

64

Manu Romero

@mrm8488

7 months

We have been working on it for the last few months!

Maisa

@maisaAI_

7 months

Introducing Maisa KPU: The next leap in AI reasoning capabilities. The Knowledge Processing Unit is a Reasoning System for LLMs that leverages all their reasoning power and overcomes their intrinsic limitations.

61

169

984

4

5

63

Manu Romero

@mrm8488

2 years

I fine-tuned @StabilityAI SD🎨on a logos + captions dataset and love some results! - 🔡Prompt: "Artificial Intelligence logo" 🤖 📚Resources: - Fine-tuning script: - Dataset: - @huggingface Space: WIP!

6

9

64

Manu Romero

@mrm8488

5 years

I made a Colab: Getting started with the AS SIMPLE AS POWERFUL @huggingface Transformers PIPELINES! #NLU #NLP #DL

Huggingface_pipelines_demo.ipynb

Run, share, and edit Python notebooks

colab.research.google.com

2

15

63

Manu Romero

@mrm8488

3 years

Zero-shot ⚡ Paraphrasing 🦜 (among other tasks) is now possible in the @huggingface model hub with @BigscienceW T0 model!

1

17

62

Manu Romero

@mrm8488

5 years

So, finally I did it! I fine-tuned #BETO ( #Spanish #BERT ) on kind of Spanish SQUAD V2 and everything was so easy thanks to @huggingface transformers API/tools. #NLU

4

13

60

Manu Romero

@mrm8488

3 years

One of the smallest models in the @huggingface for the #summarization task: Model size = 400 MB ROUGE-2 = 16.90 #NLP

3

15

61

Manu Romero

@mrm8488

1 year

🏆5️⃣0️⃣0️⃣🏆 Thank you @huggingface

6

7

58

Manu Romero

@mrm8488

1 year

Dolly + BLOOMz = #DOLLOOM coming this weekend...

4

12

61

Manu Romero

@mrm8488

2 years

Hola a todos! Hoy estaré dando un taller acerca de cómo hacer "fine-tuning" a modelos de lenguaje de gran tamaño ( #LLMs ) gracias a las últimas técnicas de de #PEFT . Entre otras cosas aprenderemos a crear nuestro propio #ChatGPT !

4

12

59

Manu Romero

@mrm8488

2 years

I fine-tuned FLAN T5 (base) 🍮 to solve Grade School Math problems 🧮 (GSM8K). I am experimenting with the large version, too, and will release the colab notebook 📒 ASAP.

4

58

Manu Romero

@mrm8488

3 years

Many people asked me about fine-tuning a model (T5) to translate from SQL to Natural language and vice versa. 🗣️↔️💽I think it is time to create a notebook 📒

2

5

58

Manu Romero

@mrm8488

1 year

Live the new "collections" new @huggingface hub feature

Spanish Language Models - a mrm8488 Collection

huggingface.co

2

21

57

Manu Romero

@mrm8488

3 years

Proud of being joined at @NarrativaAI as NLP/NLG Engineer!

11

1

57

Manu Romero

@mrm8488

4 years

My 'present' to @huggingface for being awarded as Best Demo Paper at #emnlp2020 . mT5-small fine-tuned on tydiQA for multilingual QA. EM = 41.65

4

9

58

Manu Romero

@mrm8488

1 year

Thanks "GPU-Poor" guys for the optimization/quantization techniques

1

5

55

Manu Romero

@mrm8488

3 years

A side project got 120⭐️on github:

GitHub - mrm8488/shared_colab_notebooks: A Repo to store the Google Colaboratory Notebooks that I...

A Repo to store the Google Colaboratory Notebooks that I have created and shared - mrm8488/shared_colab_notebooks

github.com

1

9

58

Manu Romero

@mrm8488

1 year

At @ClibrainAI Labs 🧪, we are pushing @huggingface #IDEFICS 🤗🦩 to its limit and testing it for creating Python snippets 👩‍💻 from a flowchart 📈! (Yes, it also works in Spanish) More news coming soon...

3

12

56

Manu Romero

@mrm8488

5 years

To finish a great week I am releasing a fine-tuned on #Spanish SquadV2 with *distillation* model! Using the same teacher as #DistilmBert . Thanks again to @huggingface Transformers for making it so easy. #NLP #NLU

1

12

56

Manu Romero

@mrm8488

6 months

It was fast! ⚡ #GaLoRE is already integrated on @huggingface Transformers✨: cc: @younesbelkada 👩‍💻code example:

1

14

56