Semanlink - Fine-tuning AND Unsupervised machine learning

> the paper LoRA: Low-Rank Adaptation of Large Language Models proposes to decompose the weight changes, ΔW, into a lower-rank representation. (To be technically correct, LoRA does not decompose the matrices directly, but it learns the decomposed matrices via backpropagation).
>
> suppose ΔW is the weight update for an A × B weight matrix. Then, we can decompose the weight update matrix into two smaller matrices: ΔW = WA WB, where WA is an an A × r-dimensional matrix, and WB is an an r × B-dimensional matrix.

LORA and LLama :

> Lit-LLaMA repository a simple, readable reimplementation of Meta’s popular LLaMA model. Besides code for training and running LLaMA itself (with the original Meta LLaMA weights), it also contains code for finetuning LLaMA using LLaMA-Adapter and LoRA.

2023-07-27 A propos

What is low-rank adaptation (LoRA)? - TechTalks

Tags:

2023-07-27 A propos

LoRA for semantic similarity tasks

Tags:

2023-07-21 A propos

clem 🤗 sur Twitter : "Llama 2 by @Meta is already integrated with @huggingface transformers, TGI, inference endpoints, PEFT and much more..."

Tags:

2023-07-19 A propos

Sanjeev Arora sur Twitter : "new `skills' induced by LLM fine-tuning can be localized in tiny fraction of the model."

Tags:

2023-07-07 A propos

[2305.12517] Retrieving Texts based on Abstract Descriptions

Tags:

2023-06-15 A propos

[2306.07536] TART: A plug-and-play Transformer module for task-agnostic reasoning

Tags:

2023-06-15 A propos

Sanjeev Arora sur Twitter : "Fine-tuning language models using just forward pass!...r

Tags:

2023-06-09 A propos

How to use Alpaca-LoRA to fine-tune a model like ChatGPT

Tags:

2023-05-26 A propos

Hyung Won Chung sur Twitter : "RLHF as an instance of using a learned objective function"

Tags:

2023-05-18 A propos

Google "We Have No Moat, And Neither Does OpenAI"

Tags:

> low-cost public involvement was enabled
by a vastly cheaper mechanism for fine tuning called low
rank adaptation ()[LoRA](tag:lora)

> **Part of what makes LoRA so effective is that ... it’s stackable.**
>
> By contrast, training giant models from scratch not only
throws away the pretraining, but also any iterative
improvements that have been made on top.

> LoRA updates are very cheap to produce (~$100) for the
most popular model sizes.

> Many of these projects are saving time by training on
small, highly curated datasets...
> These
datasets are built using synthetic methods (e.g. filtering
the best responses from an existing model) and
scavenging from other projects

> Directly Competing With Open Source
Is a Losing Proposition

> Paradoxically, the one clear winner in all of this is Meta.
Because the leaked model was theirs ([LLaMA](tag:llama)), they have
effectively garnered an entire planet's worth of free labor.
Since most open source innovation is happening on top of
their architecture, there is nothing stopping them from
directly incorporating it into their products.

2023-05-04 A propos

Andrej Karpathy sur Twitter : "There's a chance that LoRA finetunes work so well that it dramatically alters the finetuning vs. retrieval + few-shot prompting power dynamic..."

Tags:

2023-04-21 A propos

Domain Adaptation with Generative Pseudo-Labeling (GPL) | Pinecone

Tags:

2023-04-09 A propos

Daniel Vila Suero sur Twitter : "Data quality is key for LLMs, but we're building Open Source LLMs with data of "unknown" quality... Introducing Alpaca GarbageCollector..."

Tags:

2023-04-05 A propos

Alpaca Finetuning of Llama on a 24G Consumer GPU

Tags:

2023-03-27 A propos

tloen/alpaca-lora: Instruct-tune LLaMA on consumer hardware

Tags:

2023-03-22 A propos

[2106.09685] LoRA: Low-Rank Adaptation of Large Language Models

Tags:

2023-03-21 A propos

LLM Zoo at Home: LLaMA & Alpaca | bergis universe of software, hardware and ideas

Tags:

2023-03-20 A propos

Shayne Longpre sur Twitter : "A 🧵 on @OpenAI LLM "Alignment" (e.g. #ChatGPT)..."

Tags:

2023-02-27 A propos

[2212.09741] One Embedder, Any Task: Instruction-Finetuned Text Embeddings

Tags:

2023-02-17 A propos

Prompting, Instruction Finetuning, and RLHF (CS224N)

Tags:

2023-02-16 A propos

(((ل()(ل() 'yoav))))👾 sur Twitter : "there is this genre of papers that show you can train/tune only some subset of a network's weights, freezing the rest, and things still work as well as (or better than) full training/tuning..."

Tags:

2023-02-15 A propos

[2302.04870] Offsite-Tuning: Transfer Learning without Full Model

Tags:

2023-02-11 A propos

Parameter-Efficient Fine-Tuning using 🤗 PEFT

Tags:

Parameter-efficient fine-tuning (PEFT)

2023-02-10 A propos

The Flan Collection: Advancing open source methods for instruction tuning – Google AI Blog

Tags:

2023-02-02 A propos

karpathy/nanoGPT: The simplest, fastest repository for training/finetuning medium-sized GPTs.

Tags:

2023-02-02 A propos

Shayne Longpre sur Twitter : "What’s the best completely public competitor to #ChatGPT? Flan-T5 beats all public models we tested..."

Tags:

2023-02-01 A propos

[2002.01808] K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters

Tags:

2023-01-12 A propos

[2205.12410] AdaMix: Mixture-of-Adaptations for Parameter-efficient Model Tuning

Tags:

2022-12-16 A propos

[2205.05638] Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learning

Tags:

2022-12-15 A propos

ValueError "invalid literal for int() with base 10" in trainer.evaluate (dataset created from pandas) · Issue #228 · huggingface/setfit

Tags:

2022-12-13 A propos

Talking to Models: Stanford U & Microsoft Method Enables Developers to Correct Model Bugs via Natural Language Patches | Synced

Tags:

2022-11-27 A propos

Lewis Tunstall sur Twitter : "The SetFit library for few-shot learning with Sentence Transformers now supports *multi-label text classification*..."

Tags:

2022-10-14 A propos

Sanjeev Arora sur Twitter : "A priori, fine-tuning a huge LM on a few datapoints could lead to catastrophic overfitting. So why doesn’t it? Our theory + experiments..."

Tags:

2022-10-14 A propos

huggingface/setfit: Efficient few-shot learning with Sentence Transformers

Tags:

2022-10-12 A propos

[2205.11498] Domain Adaptation for Memory-Efficient Dense Retrieval

Tags:

2022-09-26 A propos

[2209.11055] Efficient Few-Shot Learning Without Prompts

Tags:

2022-09-23 A propos

[2209.00099] Efficient Methods for Natural Language Processing: A Survey

Tags:

2022-09-04 A propos

Recent Advances in Language Model Fine-tuning (Feb 2021)

Tags:

2022-09-02 A propos

[2010.07835] Fine-Tuning Pre-trained Language Model with Weak Supervision: A Contrastive-Regularized Self-Training Approach

Tags:

2022-09-02 A propos

[2106.10199] BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-models

Tags:

2022-09-01 A propos

On Stability of Few-Sample Transformer Fine-Tuning | Kaggle

Tags:

2022-08-29 A propos

Unsupervised Learning — Sentence-Transformers documentation

Tags:

2022-08-20 A propos

AdapterHub: A Framework for Adapting Transformers | Towards Data Science

Tags:

Adapter modules (LM finetuning)

2022-07-22 A propos

adapter-hub/adapter-transformers: Huggingface Transformers + Adapters

Tags:

2022-07-22 A propos

Domain transfer with GGPL: German Generative Pseudo Labeling 🥨 | by Matthias Richter | Jun, 2022 | ML6team

Tags:

2022-06-02 A propos

Nils Reimers sur Twitter : "GPL goes multi-lingual..."

Tags:

2022-06-01 A propos

Ramsri Goutham Golla sur Twitter : "Hi @Nils_Reimers For GPL you used "msmarco-distilbert-base-tas-b" model and ..."

Tags:

2022-04-27 A propos

[2008.11228] A simple method for domain adaptation of sentence embeddings

Tags:

2022-04-01 A propos

[2004.05119] Beyond Fine-tuning: Few-Sample Sentence Embedding Transfer

Tags:

2022-03-31 A propos

Sentence Transformer Fine-Tuning (SetFit): Outperforming GPT-3 on few-shot Text-Classification while being 1600 times smaller | by Moshe Wasserblat (2021-12)

Tags:

2022-03-31 A propos

Nils Reimers sur Twitter : "Great post on SetFit"

Tags:

2022-03-31 A propos

Sentence Embedding Fine-tuning for the French Language | by La Javaness R&D | Feb, 2022 | Medium

Tags:

2022-03-31 A propos

Domain Adaptation — Sentence-Transformers documentation

Tags:

2022-03-31 A propos

[2203.14655] Few-Shot Learning with Siamese Networks and Label Tuning

Tags:

> the problem of building text classifiers with little or no training data.
>
> In recent years, an approach based on neural textual entailment models has been found to give strong results on a diverse range of tasks.

(cf. #[NLI](tag:nli), using the input text as the premise and the text representing the label as the hypothesis)

> In this work, we show that **with proper pre-training, Siamese Networks that embed texts and labels** offer a competitive alternative.
>
> We introduce **label tuning: fine-tuning the label embeddings only**. While giving lower performance than model fine-tuning (which updates all params of the model), this approach has the architectural advantage that a single encoder can be shared by many different tasks (we only fine-tune the label embeddings)
> The drop in quality can
be compensated by using a variant of **[Knowledge distillation](tag:knowledge_distillation)**

[Github](https://tinyurl.com/label-tuning), [Tweet](doc:2022/03/thomas_muller_sur_twitter_pa)

2022-03-30 A propos

Domain adaptation of word embeddings through the exploitation of in-domain corpora and knowledge bases (PhD Thesis 2021)

Tags:

Thèse de Hicham El Boukkouri, univ Paris-Saclay

[Github](https://github.com/helboukkouri/phd-code)

### Goal

Given a target specialized domain, improve the quality of general-domain
word representations using in-domain corpora and/or knowledge bases

### Contributions

#### a method for specializing general-domain embeddings in a [Low-Resource](tag:nlp_low_resource_scenarios) context.

> - train static representations on the task corpus, 
> - resume the
pre-training of general-domain contextual embeddings on the same task corpus,
> - finally, combine both static and contextual representations into
one final model

#### we tackle the issue of using a general-domain vocabulary in a specialized domain

#### Evaluation of  re-training vs training from scratch on specialized corpora using a specialized vocabulary

training from scratch better, but not that much: re-training from a general model
is still appropriate as it is less expensive and leads to comparable, although
slightly lower, performance

#### Regarding subword-based tokenization systems
> we argue that they are inconvenient in practice -> CharacterBERT, a variant
of BERT that uses ELMo’s character-based system instead of WordPieces. More convenient ti use, superior robustness to misspellings

#### Ways to specialize general-domain representations using knowledge bases

a strong baseline using a simple
method relying on graph embeddings and concatenation, using only is_a relation

> both static and contextual embeddings may effectively be
specialized using this simple approach

#### Knowledge Injection Modules (KIM) that
inject the knowledge representations directly within the BERT-like models' architecture

### Notes

> our experiments focused on a single
setting (i.e. the medical domain and the English language)

> meta-embeddings, an approach that consists in combining
different sets of representations for achieving improved performance

2022-03-23 A propos

[2006.05987] Revisiting Few-sample BERT Fine-tuning

Tags:

2022-03-21 A propos

NAVER LABS Europe : "@Nils_Reimers of @huggingface on 'Unsupervised domain adaptation for neural search'"

Tags:

2022-03-09 A propos

[2112.07577] GPL: Generative Pseudo Labeling for Unsupervised Domain Adaptation of Dense Retrieval

Tags:

An unsupervised domain adaptation technique for dense retrieval models

1. synthetic queries
are generated for each passage from the target corpus (using an existing pre-trained [T5](tag:text_to_text_transfer_transformer)
encoder-decoder)
2. the generated queries are used for mining negative
passages (retrieving the most similar
paragraphs using an existing dense retrieval
model == hard negatives!)
3. the query-passage pairs are labeled by a cross-encoder and used to train the domain-adapted
dense retriever (using method described in [Hofstätter et al.,
2020](doc:2021/12/2010_02666_improving_efficien))

[Nils Reimers sur Twitter](doc:2021/12/nils_reimers_sur_twitter_do_), [GitHub](https://github.com/UKPLab/gpl),  by the author of [TSDAE](doc:2021/09/2104_06979_tsdae_using_trans)

Claims to improve "Doc2Query" [Document Expansion by Query Prediction](doc:2022/01/1904_08375_document_expansion): ([src](https://twitter.com/KexinWang2049/status/1471435779415150598))

> - GPL: Uses doc2query to construct synthetic data and does knowledge distillation (i.e. training) on that data.
> - Doc2query: Generates queries to extend the documents and use BM25 on top of them w/o training.

2021-12-15 A propos

Unsupervised Training for Sentence Transformers | Pinecone

Tags:

2021-11-24 A propos

On the Stability of Fine-tuning BERT: Misconceptions, Explanations, and Strong Baselines (2021)

Tags:

2021-10-30 A propos

[1908.11860] Adapt or Get Left Behind: Domain Adaptation through BERT Language Model Finetuning for Aspect-Target Sentiment Classification

Tags:

2021-10-21 A propos

[2106.04647] Compacter: Efficient Low-Rank Hypercomplex Adapter Layers

Tags:

2021-09-29 A propos

[2104.06979] TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning

Tags:

> The most
successful previous approaches like InferSent (Conneau
et al., 2017), Universial Sentence Encoder
(USE) (Cer et al., 2018) and SBERT (Reimers and
Gurevych, 2019) heavily relied on labeled data to
train sentence embedding models.
>
> TSDAE can
achieve up to 93.1% of the performance of indomain
supervised approaches. Further, we
show that TSDAE is **a strong domain adaptation
and pre-training method for sentence
embeddings**, significantly outperforming other
approaches like Masked Language Model.

> During training, TSDAE
encodes corrupted sentences into fixed-sized
vectors and requires the decoder to reconstruct the
original sentences from this sentence embedding.

- <https://www.sbert.net/examples/unsupervised_learning/TSDAE/README.html>
- [github](https://github.com/UKPLab/sentence-transformers/tree/master/examples/unsupervised_learning/TSDAE)
- [UKPLab/sentence-transformers: Sentence Embeddings with BERT & XLNet](doc:2020/07/ukplab_sentence_transformers_s)
- [twitter](https://twitter.com/KexinWang2049/status/1433361957579538432):

> **TSDAE can learn domain-specific sentence embeddings with unlabeled sentences**
>
> Most importantly, instead of STS (Semantic Textual Similarity), **we suggest evaluating unsupervised sentence embeddings on the domain-specific tasks&datasets, which is the real use case for them**. Actually, STS scores do not correlate with performance on specific tasks.

2021-09-01 A propos

[1902.00751] Parameter-Efficient Transfer Learning for NLP

Tags:

2021-04-11 A propos

How many data points is a prompt worth?

Tags:

2021-04-08 A propos

Recent Advances in Language Model Fine-tuning

Tags:

2021-02-24 A propos

[2004.10964] Don't Stop Pretraining: Adapt Language Models to Domains and Tasks

Tags:

2020-12-01 A propos