Gradient
@Gradient_AI_
Followers
4,156
Following
45
Media
235
Statuses
391
Advancing AI Agents for the Modern Enterprise.
Joined August 2023
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Adalet
• 192561 Tweets
Linkin Park
• 188537 Tweets
Gazze
• 117684 Tweets
Engin Polat
• 113053 Tweets
ベイマックス
• 102148 Tweets
Chester
• 96814 Tweets
雇用統計
• 37124 Tweets
Game Day
• 36623 Tweets
大阪府警
• 35838 Tweets
NANON LETSGO NYFW
• 35768 Tweets
エイリアン
• 33655 Tweets
#サクサクヒムヒム
• 31946 Tweets
#ساعه_استجابه
• 30476 Tweets
金メダル
• 29204 Tweets
#浦島坂田船11周年
• 22857 Tweets
#DAY6_9th_Page
• 22276 Tweets
KIDNAP CHAPTER ONE
• 21550 Tweets
Franco Escamilla
• 20680 Tweets
HAPPY BIRTHDAY TANNIE
• 20408 Tweets
#데이식스_청춘의_9번째_페이지
• 13850 Tweets
おぱんちゅうさぎ
• 11552 Tweets
ロムルス
• 11407 Tweets
Pinned Tweet
We’re going back 2 back! 🔥 Introducing the first 1M context window
@AIatMeta
Llama-3 70B to pair with the our Llama-3 8B model that we launched last week on
@huggingface
. Our 1M context window 70B model landed a perfect score on NIAH and we’re excited about the results that
40
112
659
We've been in the kitchen cooking 🔥 Excited to release the first
@AIatMeta
LLama-3 8B with a context length of over 1M on
@huggingface
- coming off of the 160K context length model we released on Friday!
A huge thank you to
@CrusoeEnergy
for sponsoring the compute. Let us know
67
260
1K
4M Context Length Llama-3 8B (V0.1) 🧑🍳
We’ve been actively pushing the boundary of long context models, which is why we’re excited to share an early version of our 4M context window Llama-3 8B on
@huggingface
. While perfecting a recipe takes time, this marks the longest
23
69
460
We just released the first LLama-3 8B with a context length of over 160K onto Hugging Face! SOTA LLMs can learn to operate on long context with minimal training (< 200M tokens, powered by
@CrusoeEnergy
's compute) by appropriately adjusting RoPE theta.
🔗
31
78
421
Are you at AIAI Summit hosted by
@AI_AInstitute
in San Jose? If you’re around, swing by our booth and come say hi! 👋
67
9
227
We just released our API for RAG! Now you can setup RAG using Gradient's SDK or web UI. Try it out for yourself!
🔗
✅ Setup RAG in Seconds
✅ Powered by MongoDB and LlamaIndex
✅ Fully Optimized Out-of-the-Box
64
4
212
It's Friday night and our kitchen’s still open! 🍳 Now serving a 524K context window
@AIatMeta
Llama-3 70b on
@huggingface
, which has been trained on a more extensive proprietary chat dataset to give the model chat ability over long sequences! The best part is that we’re still
17
40
287
Love this quick tutorial and breakdown from
@GregKamradt
on RAG and how easy it is to use Gradient's managed RAG service!
✅ Production-Grade RAG in Seconds
✅ Few Lines of Code
✅ Powered by
@llama_index
and
@MongoDB
61
1
207
Join us for our final episode of our 3 part series around Time Series AI. Hear from our Gradient AI Research Award recipients from
@CityUHongKong
and their research on the FTS-Diffusion framework - a novel generative model that addresses the complexities of financial time series
61
3
202
Don't worry, we didn't forget about 70b🥳 Take a look at the first
@AIatMeta
Llama-3 70b model with a context length of 262K - scoring a perfect retrieval for NIAH! We included an extensive proprietary chat dataset to give the model chat ability over long sequences as well.
A
17
46
228
Lately we've been exploring several open-source frameworks, in addition to building our own in-house abstractions to create more robust AI systems. Take a look at our learnings and deep dive into DSPy.
🔗
✅ GPT-4 Performance at 10x Lower Cost
✅
1
28
130
Did you catch our breakdown and deep dive into
#DSPy
? Take a look at how you can achieve GPT-4 performance for 10x less the cost, improve accuracy, and reduce the response size.
🔗
2
24
123
With the recent release of our 1M context length Llama-3 8B model, there's been a lot of great conversation around evals. Here's our take on one of the most popular evals - Needle in a Haystack.
🔗
✅ Why Haystack Diversity Matters
✅ Impact of Haystack
2
17
97
We're excited to announce our partnership with
@MongoDB
. See how you can easily create a financial chatbot with MongoDB and
@llama_index
. Learn how to use:
😵💫 RAG to Reduce Hallucination
🍃 Mongo for Vector Store
🦙 LlamaIndex for RAG Orchestration
0
22
94
Recently we've launched a 4M Context Length Llama-3 8B and a 1M Context Length Llama-3 70B from
@AIatMeta
. Take a look at why our team is now adopting RULER to evaluate these models vs. using just NIAH.
🔗
✅ Our model is behind only GPT-4 and Yi in
2
15
81
NIAH is not representative of how a long context model will perform in practical use cases, which is why we took it a step further and evaluated our 1M context length
@AIatMeta
Llama-3-8B using RULER to get a more comprehensive view on quality.
✅ Our model is behind only GPT-4
3
10
70
In order to increase the context window to 4M on
@AIatMeta
’s Llama-3 model, one of the things we had to do was increase RoPE theta and add in full context length supervised fine-tuning. Take a peak as some of our learnings as we dive into:
🔗
✅
0
18
70
Need inference for long context models (262k-1M+)? Join our custom agent and long context waitlist:
And in the meantime, you can find community discussions and related announcements on our Discord:
3
3
49
Nvidia GTC has been amazing. Shout out to the
@nvidia
team for putting on an amazing event and thank you to everyone who stopped by to chat! Glad everyone enjoyed our “old school machine”!
17
11
42
We're excited about how our 1M Context Length
@AIatMeta
Llama-3 70B performed against RULER. Dive into the results or checkout how we made this possible (e.g. scaling rotational embeddings).
✅ RULER Results:
✅ Made Possible By:
0
13
38
The results are in! As we continue to invest into how we evaluate long context models, take a look at how Gradient’s 1M Context Length
@AIatMeta
Llama-3 70B performed against RULER.
🔗
A huge thank you to
@CrusoeEnergy
for the continued partnership and
0
7
32
Excited that the RULER Benchmark now includes Gradient's 1M Context Length
@AIatMeta
Llama-3 70B, placing it 4th overall! 🙌
Take a look at some our deep dives into long context to learn more:
✅ RULER vs. 1M Llama3 70B:
✅ RULER
2
4
29
We’re excited to announce that Llama 3 is now available on Gradient! Take Llama 3 for a spin, without having to worry about the setup or infrastructure. A huge congrats to the team at
@Meta
for all the hard work that they put into this release! Here’s a quick breakdown of Llama
0
9
23
We just released a 1M context window Llama-3 70b! Check it out:
We’re going back 2 back! 🔥 Introducing the first 1M context window
@AIatMeta
Llama-3 70B to pair with the our Llama-3 8B model that we launched last week on
@huggingface
. Our 1M context window 70B model landed a perfect score on NIAH and we’re excited about the results that
40
112
659
1
4
23
We had a great time this morning chatting about domain-specific LLMs at
@genaisummitsf
. If you want to learn more about domain-specific models in enterprise, check out our recent post or visit us at our booth!
🔗
📍 Booth # B8
Shoutout to the other
0
5
23
If you just stumbled on this post, we now have a 524K model available 😉
It's Friday night and our kitchen’s still open! 🍳 Now serving a 524K context window
@AIatMeta
Llama-3 70b on
@huggingface
, which has been trained on a more extensive proprietary chat dataset to give the model chat ability over long sequences! The best part is that we’re still
17
40
287
2
4
23
Discover the fastest way to setup RAG, using our newest Accelerator Block - powered by
@llama_index
and
@MongoDB
.
🔗
🏎️ Setup RAG in Seconds
✅ Production Grade & Fully Managed
👍🏽 Optimized RAG Performance Out-of-the-box
📞 Accessible Via an API Call
0
9
22
Considering which LLM you should use? Take a look at the rise of domain-specific models in enterprise.
“By 2027 more than 50% of GenAI models that are used by enterprise will be domain-specific - focusing on either an industry or business function.” -
0
2
21
In case you haven't seen our post, we just released a 1M context window Llama-3 70B!
We’re going back 2 back! 🔥 Introducing the first 1M context window
@AIatMeta
Llama-3 70B to pair with the our Llama-3 8B model that we launched last week on
@huggingface
. Our 1M context window 70B model landed a perfect score on NIAH and we’re excited about the results that
40
112
659
1
2
22
Gradient's Co-Founder and Chief Architect,
@markatgradient
, will be speaking in
@MavenHQ
's Lightning Lesson series about AI Agents and Multimodal LLMs. Join us to learn about:
✅ Why and When to Use Task Adapted Models
✅ How to Fine-tune LLMs for Task Adaptation
✅ How You Can
0
4
20
The major challenge in training long context models isn't just improving attention implementation. It's also the scarcity of adequate long context training data. In order train our
@AIatMeta
Llama-3 long context models, we developed a synthetic data generation pipeline that
0
5
18
Gradient's Chief Scientist,
@LeoPekelis
, recently sat down with
@VentureBeat
to discuss how we developed an open-source
@AIatMeta
Llama-3 model with a context length of 1M+ 🙌
As always, a huge thank you to
@CrusoeAI
for sponsoring the compute!
1
6
19
We had a great time at
@ai_expo
this year! Thank you to everyone who attended
@markatgradient
's session and stopped by our booth to say hi 👋 For those who weren't able to make it, here's a look at some of our innovative AI-driven solutions across various industries.
💰 FinServ:
0
3
19
If you haven't seen our interview w/
@cerebral_valley
, check out how our Co-Founder and CEO,
@chrisatgradient
, is thinking about the future of enterprise.
✅ Impact of Agents in Enterprise
✅ Industry Needs & Nuances w/ AI
✅ Gradient 12 Months From Now
1
6
18
Appreciate everyone who attended
@markatgradient
's session this morning on solving complex enterprise use cases with agents at AIAI Summit. It was great meeting everyone and please stop by the booth if you haven't yet to say hi! 👋
Thank you
@AI_AInstitute
for an awesome
0
4
17
📣 It’s Launch Day! 📣
We’ve just released the Gradient API to help you build private LLMs that you own. We simplify inference and fine-tuning on open-source LLMs (i.e., llama2). 🚀🦙
Sign up at and check us out on
@ProductHunt
!
2
4
17
There's some talented creators out there.
@GregKamradt
uses Gradient's fine-tuning capabilities to clone himself in latest video. 😲 Why Gradient?
✅ No need for dedicated infrastructure
✅ Cost Effective
✅ Integrations w/ partners Like
@LangChainAI
0
7
16
Our Co-Founder and CEO,
@chrisatgradient
sat down with
@cerebral_valley
to chat about the future of enterprise AI.
✅ The Founding Premise of Gradient
✅ Why Agents are the Future of Enterprise AI
✅ Gradient's Goals for the Next 12 Months
0
8
16
Take a look at why we had to implement a new pipeline for synthetic long context data generation in order to overcome challenges in training our 1M+
@AIatMeta
Llama-3 context models 🧐
🔗
✅ Shifting from Pre to Post-Training
✅ Why Synthetic Data is
0
3
16
Our CEO and Co-Founder,
@chrisatgradient
, will be taking the stage at AI in Fintech Forum 2024 hosted by
@Stanford
. Come join us as Chris shares his thoughts with other leaders on all things AI and where we're heading when it comes to GenAI for Financial Services.
🔗
0
5
15
Thank you
@Stanford
for having us at this year’s AI in Fintech Forum! Our CEO,
@chrisatgradient
, had an amazing time sharing his thoughts with other thought leaders on where the financial services industry is headed with AI 💰📈💵
0
1
14
We're headed to
@ai_expo
this week! Come find us at our booth or at our session, where we dive into how you can solve enterprise use cases using AI agents!
🔗
📍 Santa Clara Convention Center
📅 June 5th - 6th
👋 Booth # 85 (Intelligent Automation)
0
1
14
Discover how we worked with an asset manager to help automate and enhance their trade settlement process using Gradient Agent Foundry.
#AIAutomation
🔗
📉 75% ↓ in Manual Processing
📉 85% ↓ in Error Rates
📈 Improved Regulatory Compliance
0
3
14
We're excited to present
@hayou_soufiane
with Gradient's AI Research Fellowship. We cannot wait to see the impact that he’ll have as he zeros in on optimizing scaling methods for fine-tuning LLMs and using Low Rank Adaptation. Congratulations!
Learn More:
0
4
14
We're excited to announce our new integration with
@Haystack_AI
by
@deepset_ai
- working directly with Haystack 2.0.
✅ Generate Vector Embeddings in Haystack pipelines
✅ Use Gradient’s Models w/ Haystack pipelines for RAG and desired tasks
Learn More:
0
2
13
Our Chief Scientist,
@LeoPekelis
, sat down with
@MatthewBerman
for a fireside chat 🔥 on our 1M Context Length
@AIatMeta
Llama-3 models:
🔗
✅ How We Did It
✅ Long Context Use Cases
✅ Challenges w/ Development
✅ Performance & Evals (s/o
@GregKamradt
)
0
4
13
We are partnering with
@SamsungNext
for the Samsung Next 2024 GenAI Hackathon. Apply now and join us!
Finalists will have the chance to pitch their ideas to a world class panel of AI investors at a live Demo Day in NYC on May 11, 2024.
0
2
12
Hacking has commenced at
@MongoDB
’s Vector Search Hacakathon at AGI House! Amazing partners in attendance and we’re excited to see what they build with Gradient’s Accelerator Blocks!
0
1
13
In November, we demo'd how you can build a financial chatbot using Gradient,
@llama_index
, and
@MongoDB
. Learn how to recreate this very same financial chatbot 10x faster without sacrificing quality or performance, using Gradient Accelerator Blocks.
🔗
1
3
13
We have a busy week with some amazing topics to discuss across all things AI!🔥
✅ Long Context Windows w/
@AIMakerspace
:
✅ Building Agentic Workflows w/
@MavenHQ
:
✅ GaLore & InRank w/
@jiawzhao
:
0
3
13
Our team had a fantastic time at
@TensorWaveCloud
's fireside chat last night to discuss the future of compute. Thank you
@DarrickHorton
and team for letting
@LeoPekelis
join in on the fun 🙏
Cannot wait to join in on the next one. If you're interested in how Gradient is
2
2
13
We'll be chatting with
@MononitoGoswami
on Wed. April 3rd at 10am PST for part 2 of 3 in our livestream series on Time Series AI. Register today so you don't miss out and learn more about his research around the MOMENT Model! 🙌
1
5
12
Join us on Tuesday June 25th as Gradient’s Co-Founder and Chief Architect,
@markatgradient
, will be joining
@TensorWaveCloud
and other thought leaders for an exciting fireside chat on the future of compute.
🎟️
@DarrickHorton
- CEO of TensorWave
2
4
12
If you're looking for a way to serve custom embeddings, check out this open-source framework by one of our very own engineers - Michael Feil 🙌
Thank you
@dstackai
for publishing an example on how to use this framework using a single command!
1
3
11
Did you catch our latest customer story featuring Sinpex? Take a look at how Gradient helped automate their KYB process (Know Your Business) with AI-powered data extraction.
🔗
✅ 35% Point Increase in Accuracy
✅ Inability to Respond to Q's Dropped From
0
2
12
Co-founder and Chief Architect,
@markatgradient
, was recently on
@SoftwareHuddle
to chat about autonomous AI agents and why it's been such an exciting area to work on. Check out the full episode here:
Thank you
@seanfalconer
&
@alexbdebrie
for having us!
7
4
12
We’re excited to share our open-source framework that
we worked with
@awscloud
that lest you score different generative language models across various evaluation tasks and benchmarks - used by leaderboards such as Hugging Face!
1
4
10
2
1
12
We've been cooking 🧑🍳! 1M model now available on Hugging Face:
0
2
11
Did you miss our chat yesterday with
@MononitoGoswami
on the MOMENT model? If so no worries - you can check out the recording to see what you missed out on. Stay tuned for part 3 of our livestream on Time Series AI.
Thank you Mononito for joining us! 🙌
0
4
11
Join us on Tues. May 14th @ 10am PST as
@markatgradient
chops it up with
@yujian_tang
on Tuesday Tech Talks!
🔗
✅ Model Innovation & Open Source Community
✅ Challenges & Complexities of Long Context Models
✅ Performance and Eval Considerations
✅ Where
0
2
11
We’re here at the
@MongoDB
Generative AI Breakfast to share our thoughts around all things AI!
If you’re here at AWS Re-Invent, let us know! We would love to connect.
1
4
11
@KeldenLin
@CrusoeEnergy
More checkpoints are coming! We are still working but wanted to get something out for the community in the interim :)
0
0
12
Did you catch our announcement this week? We launched our AI Development Lab to offer enterprise organizations a white glove, end-to-end development service to help build private, custom LLMs and AI copilots!
Learn More:
0
2
11
We had a fantastic time at
@SapphireVC
’s 3rd Annual Hypergrowth Engineering Summit! 🎉✨ Thank you to everyone who attended
@markatgradient
's session on Long Context Agents.
Interested in how long context agents can help automate your enterprise business? Lets chat!
0
1
11
Gradient’s Healthcare LLM, Nightingale, has helped healthcare organizations build AI-driven solutions over the last few months - expediting manual tasks and drastically lowering overhead costs.
✏️ Clinical Transcriptions:
🏥 Healthcare Data Management:
0
2
11
We just released the newest addition to our Agent Toolkit - Audio Transcription. Simply upload the file you want to transcribe and our audio transcription API will return a near perfect extraction of the text from the audio.
Learn More:
1
3
11
When we build domain-specific models like Albatross, we place responsible AI top of mind. Take a look at how Gradient focuses on the following four pillars for responsible AI:
✅ Logic & Reasoning
✅ Privacy
✅ Security
✅ Regulation & Compliance
0
3
10
Have you checked out our Alphatross LLM yet on Hugging Face? Alphatross is an early version of our Albatross model with limited capabilities, specializing in financial services. However It has the highest H6 rating across Llama2-70B variations!
🔗
0
3
11
Discover how Gradient Agent Foundry helped automate the accounts payable process for an electric vehicle manufacturer processing 30K+ invoices.
🔗
📉 ↓ Workload by ~80%
📉 ↓ Error Rate by 65%
🚀 ↑ Processing Times by 70%
💰 Cost Savings of ~60%
0
3
11
What an episode! Join our Co-Founder & Chief Architect,
@markatgradient
, as he breaks down the intricacies of scaling Llama3 beyond a 1M context window on the latest episode on
@latentspacepod
!
🔗
✅ History of Long Context
✅ Intro to RoPE, ALiBi, and
0
2
11
We’re excited to welcome Jeff to our team as our newest business operations lead! Prior to Gradient, Jeff worked on a variety of roles across analytics, finance, GTM, and product at companies like Segment, Google and Opendoor. 🥳
3
1
11
What's RAG and should your enterprise business consider it? Check out our RAG 101 for Enterprise deep dive to learn about:
✅ RAG and if it's right for you
✅ How it works & how to optimize LLMs
✅ How to overcome challenges w/ Gradient &
@llama_index
0
4
11
Our Co-founder and CEO, Chris Chang, had the opportunity to sit down with
@thenewstack
to chat about all things AI.
✅ Easy to Use Development Tools
✅ Gradient’s Industry Specific LLMs
✅ AI Development Challenges in Enterprise
0
4
9
We had a great time at
@ImagineAILive
'24 in NYC! Thank you to everyone who came out and sat in on
@chrisatgradient
’s keynote and workshop on enterprise AI automation. Interested in automating your enterprise workflows? Lets chat!
0
3
10
Excited to share how a recent US bank used our domain specific finance LLM, Albatross, to power their AML program.
✅ Detected 3-4x more instances of 1st & 3rd party fraud
✅ Reduces false positives by 50%+
✅ Provided auditable & explainable outputs
0
3
10
In case you missed our post yesterday, here's our take and deep dive on the Need in a Haystack (NIAH) eval - courtesy of our Chief Scientist
@LeoPekelis
.
1
0
10
Distributed LLM Training enables users to split workloads across multiple GPUs to drastically accelerate the training process.
Take a look at how Gradient thinks about infrastructure and efficiency optimizations.
✅ Orchestration
✅ Fault Tolerance
0
3
10
Curious about Long Context Windows? Our Chief Scientist,
@LeoPekelis
, will be live on
@AIMakerspace
on Wed. May 22nd @ 10am PST to chat about:
🔗
✅ Extending Llama 3’s Context Length > 1M
✅ Long Context Evals (NIAH, RULER, etc.)
✅ RAG vs. Long-Context
0
3
10
We're excited to announce our partnership with
@SnowflakeDB
, enabling the use of Gradient through Snowpark Container Services and the release of our Gradient Datapacks on Snowflake Marketplace!
Learn More:
2
4
10
We’re headed to
@_TechSummit_
next week 🙌
@markatgradient
will be on the Data Driven Innovation Panel to discuss the future of tech and the pivotal role that data plays in ensuring its success!
📍Santa Clara Conv. Center
📅 June 19th @ 2:25pm Room 3
🎟️
0
3
10
Solving for AML? Albatross, Gradient’s proprietary domain-specific model for finance, has already helped a large US bank enhance their transaction monitoring processes to detect suspicious activities more efficiently.
🔗
📈 Detect 3-4x More Instances of
0
1
10
Amazing event tonight hosted by
@useOrb
at
@Cloudflare
’s HQ. Super insightful talks from different teams and leaders in regard to ROI of GenAI innovation. 🙌
0
3
10
Have you checked out part 1 of our breakdown on how we conduct distributed LLM training at Gradient using our own proprietary distributed training platform?
✅ Orchestration
✅ Fault Tolerance
1
2
10
Please join us in welcoming our newest product lead, Brenda Lee! 🔥
Prior to joining Gradient, Brenda spent a few years leading product in the decentralized data storage space, focusing on storage protocol design and node provider software. She also spent many years at Microsoft
1
1
10
Gradient Co-Founder and Chief Architect
@markatgradient
shared our vision for building the next million AI systems at
@AIconference
this morning.
The future of AI systems is an ensemble of small "expert" models to give you the highest performing, cost-effective AI system.
3
5
10
Our team at Gradient will be heading to
@Ai4Conferences
this year! Come check our session as Gradient's Co-Founder and Chief Architect,
@markatgradient
, dives into how to use agentic systems to drive operational efficiency! 🙌
📍 Las Vegas
📅 August 12- 14th
👋 Booth
#608
0
5
9
We’ve been invited to join
@sagecodes
and the
@union_ai
team to a fireside chat! 🔥🏕️ Come join us as
@markatgradient
deep dives into down how long context models can help improve agentic workflows!
🔗
📅 June 19th @ 10am PST
0
2
9
We just released a new white paper by Gradient’s Chief Scientist, Leo Pekelis, as he dives into how AI's ROI in financial institutions.
🔗
✅ Advantages of LLMs in Financial Applications
✅ Requirements for LLM Applications in Finance
✅ And Much More...
0
6
9
We’re heading to AIAI Summit next week, hosted by
@AI_AInstitute
. Co-Founder & Chief Architect,
@markatgradient
, will be talking about how you can solve enterprise use cases through agents.
🔗
📍 San Jose Mariott
📅 Apr. 17th
🕤 9:30-10am PST
1
3
9
We're here at
@MongoDB
's DevOps for GenAI Hackathon and the hacking has started! These participants are some of the very first to use Gradient's Accelerator Blocks that was announced today, simplifying the AI development process.
Learn More:
0
3
9
Did you catch our talk earlier this month at Multimodal Weekly, hosted by
@twelve_labs
? If not, here's your chance to view the recap!
✅ Fine-Tuned Small Models > Large Models
✅ Tackling the Enterprise Data Corpus
✅ Leverage MoE & Low-Rank Adaption
1
1
9