Top Tweets for #FINETUNING
Fine-tuning search agents with multi-turn RL on Amazon SageMaker AI shows small models can match frontier model reliability at lower cost. MTRL trains full decision sequences, improving retrieval by embedding environment-specific behavior. #FineTuning #FrontierModels
@abidlabs Agent-ready Jobs docs for Transformers/TRL/Unsloth/Axolotl — hand them the pack and let the fine-tune actually start.#Agents #HuggingFace #FineTuning
Day 122 —> Becoming an AI Engineer
LLM Fine-tuning:
Classification → specialized outputs
Instruction → follows task-specific prompts
Insight: Fine-tuning doesn’t rebuild the model—it specializes existing capabilities for a target behavior.
#LLM #FineTuning #GenAI
📰 Transfer Map Predicts Which Source Tasks Hurt Instruction Tuning
Transfer between instruction-tuning tasks is signed and asymmetric; a fitted map halves prediction error and recovers up to 14...
Read more → https://t.co/4jvZp8hTam
#AIIndustry #FineTuning #AIReasoning
📰 Skin-Deep metric predicts LLM refusal fragility before fine-tuning
Across 21 models, safety-subspace geometry in residual streams predicts how fast refusal erodes under benign fine-tuning, giving teams a...
Read more → https://t.co/J68LQrJaVb
#AI #AiSafety #FineTuning #LLM
@karpathy autoresearch applies the automated agentic loop to model-training research.
My recent experiment applies it to the model fine-tuning - https://t.co/VssSUUUzjF
https://t.co/TTi38S1ZHB
#AgenticAI #AutoResearch #FineTuning #LLMEvaluation
DASA fine-tunes LLMs with synthetic embeddings, not readable text.
Authors report comparable or better results than natural-language data; in tested settings, synthesis was 3.6 to 4.9 times faster than GRADMM.
https://t.co/jx85N1bGJP #DASA #FineTuning #arXiv
Should you fine tune a Mistral model that keeps getting it wrong?
Not first. Climb in order:
1. Clearer prompt with examples
2. Bigger model
3. Fine tune, only if both failed
$4 per job is cheap. $2 a month per hosted model is what adds up.
#Mistral #FineTuning #LLM #AI
Mastering LLM Fine-Tuning with QLoRA: Quantization on a Single GPU + Code. Stop complaining about GPU costs. With QLoRA, you can fine-tune massive models on a single consumer card. so what is your excuse now? Guide and code 👉 https://t.co/HFZOo02gr6 #QLoRA #FineTuning #AI
📰 HybridCUA reaches 53.6% on OSWorld with joint GUI and CLI training
A 9B computer-use agent trained via SFT and CLI-aware RL reaches 53.6% on OSWorld, showing shell routing is a practical lever for...
Read more → https://t.co/trVOburgon
#AI #AIAgents #FineTuning #LLM
📰 KwaiMind Optimizes Product Image Editing for Marketplace CTR
Kuaishou’s KwaiMind trains on 1.8M e-commerce pairs to optimize CTR, reporting a 2.44% lift in A/B tests and introducing...
Read more → https://t.co/9mwQr9tx2U
#AI #ComputerVision #DeepLearning #FineTuning
Fine-tuning ROI goes beyond better outputs. It can create value through three areas: quality and consistency, lower inference costs, and faster user experiences.
Explore the full article - https://t.co/Ophw0EfFmo
#finetuning #mallowtechnologies
📰 Confidence fine-tuning cuts LLM reasoning tokens up to 25 percent
Confidence training on 600 problems cuts reasoning token output up to 25 percent without any length penalty or early-stopping...
Read more → https://t.co/YEmqpGijvl
#AI #FineTuning #InferenceOptimization #LLM
The fine-tune underperformed a well-routed prompt, and nobody checked the labels first.
Here is the decision order that actually works 👇
https://t.co/FFGpgtrFXm
#AI #LLM #GenerativeAI #FineTuning
Anyone hit a different failure mode? Reply below.
Fine-tuning an LLM isn't just about choosing a model. Data, team expertise, cost, and a clear business outcome all matter before training begins.
Explore the full article - https://t.co/JHrqYcrnlM
#finetuning #mallowtechnologies
Less magic, more understanding what’s happening under the hood. 🌱
#100DaysOfLearning #BuildInPublic #AIEngineering #PromptEngineering #ContextEngineering #FineTuning
Stop burning money on massive model training. You can fine-tune 235B parameter models using LORAs by adapting just 5% of the parameters. It is affordable and accessible now. Check the case studies at https://t.co/1EsK5MDKCB. #AI #LLM #FineTuning #DataScience
Only using closed AI means someone else controls your choices.
Manos Koukoumidis on AI Masterminds: open gives you the flexibility and control to build what's actually differentiated for your business.
https://t.co/rwL272GFhz
#FineTuning #OpenSourceAI
Taped with Mozilla's Raffi Krikorian.
【明日開催！InfiniStudy】
現地枠残り僅かではありますが、お待ちしております。
まだまだ飛び入り参加もOKです！
JAICON参加の方々にFine tuningやAI活用法についてお話をしていただきます
ぜひぜひご参加ください
https://t.co/ZrgLyXASFw
#InfiniStudy
#AI駆動
#Finetuning
#JAICON
Last Seen Hashtags on Sotwe
เรียลมัธยมต้น
Seen from Thailand
น้องเมีย
Seen from Thailand
adanabayan
Seen from Netherlands
benaturallyconfident
Seen from United States
schoolies
Seen from Italy
VideoChatApps
Seen from Brazil
محارم_عربية
Seen from Saudi Arabia
AZIsBlue
Seen from United States
malatyatravesti
Seen from Turkey
รับงานโคราช
Seen from Thailand
Trends for you
Most Popular Users
Elon Musk 
@elonmusk
241.8M followers
Barack Obama
@barackobama
119M followers
Cristiano Ronaldo
@cristiano
114.6M followers
Donald J. Trump
@realdonaldtrump
111.9M followers
Narendra Modi
@narendramodi
107.2M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.4M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90.1M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.3M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.3M followers
CNN
@cnn
61.8M followers
X
@x
60.7M followers