WizardLM
@WizardLM_AI
Followers
12,215
Following
613
Media
66
Statuses
297
WizardLM, WizardCoder, WizardMath
USA
Joined October 2021
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
オーロラ
• 411486 Tweets
Joost
• 306479 Tweets
CHAWARIN ON STAGE
• 225230 Tweets
FOURTH SINGING TIME
• 139169 Tweets
#ขวัญฤทัยตอนจบ
• 128335 Tweets
#不破湊3Dライブ
• 85092 Tweets
Fulham
• 71741 Tweets
自分これ
• 59703 Tweets
Man City
• 55635 Tweets
Gvardiol
• 42174 Tweets
Manchester City
• 36446 Tweets
最高のライブ
• 35754 Tweets
#العين_يوكوهاما
• 35470 Tweets
太陽フレアのせい
• 33419 Tweets
カクレンジャー
• 32690 Tweets
HeavenlyVoice WithMrC
• 31920 Tweets
#東京タワー
• 30280 Tweets
マリノス
• 27156 Tweets
Dremo
• 24672 Tweets
アルバムとライブ
• 20999 Tweets
#FULMCI
• 20446 Tweets
自分の魚
• 15953 Tweets
わたしの誕生魚
• 15074 Tweets
Sark
• 12973 Tweets
GET WELL SOON JISUNG
• 10136 Tweets
🔥🔥🔥
Introduce the newest WizardCoder 34B based on Code Llama.
✅WizardCoder-34B surpasses GPT-4, ChatGPT-3.5 and Claude-2 on HumanEval with 73.2% pass
@1
🖥️Demo:
http://47.103.63.15:50085/
🏇Model Weights:
🏇Github:
The 13B/7B…
67
408
2K
🔥Today we are announcing WizardLM-2, our next generation state-of-the-art LLM.
New family includes three cutting-edge models: WizardLM-2 8x22B, 70B, and 7B - demonstrates highly competitive performance compared to leading proprietary LLMs.
📙Release Blog:…
77
269
1K
🫡 We are sorry for that.
It’s been a while since we’ve released a model months ago😅, so we’re unfamiliar with the new release process now: We accidentally missed an item required in the model release process - toxicity testing.
We are currently completing this test quickly…
🔥Today we are announcing WizardLM-2, our next generation state-of-the-art LLM.
New family includes three cutting-edge models: WizardLM-2 8x22B, 70B, and 7B - demonstrates highly competitive performance compared to leading proprietary LLMs.
📙Release Blog:…
77
269
1K
62
70
685
🧙♀️We not only opensource the models, but also share you how we reach that!
🚀So now let's verify step-by-step to review the whole training method of WizardLM-2 together:
-------- 🧵 --------
👉Motivation First:
As the natural world's human-generated data becomes…
🔥Today we are announcing WizardLM-2, our next generation state-of-the-art LLM.
New family includes three cutting-edge models: WizardLM-2 8x22B, 70B, and 7B - demonstrates highly competitive performance compared to leading proprietary LLMs.
📙Release Blog:…
77
269
1K
22
146
683
🔥🔥🔥Incredible work!
OSS-Instruct + 🧙♂️Evol-Instruct = 🎩Magicoder
Paper:
Code:
12
105
592
🔥 Excited to open-source the code of our Evol-Instruct！
🏆It's one the most important techniques for the success of WizardLM, WizardCoder, and WizardMath.
❤️Thanks authors:
@CanXu20
,
@victorsungo_ai
,
@tangmensan
.
🏇Let's build the Evol Universe Together!
💥Code Link:…
15
114
516
🔥🔥🔥
Introduce the newest WizardCoder model !
1. Surpass Claude-Plus (+6.8), Bard (+15.3) and InstructCodeT5+ (+22.3) on HumanEval Benchmarks
2. Paper is coming, with brand-new Evol+ methods for code LLMs
Github:
HF Weight:
12
95
460
🔥🔥🔥Introduce the newest WizardLM models trained from Llama-2 🎉 with brand-new Evol+ methods.
WizardLM-13B-V1.2 achieves:
(1). 7.06 on MT-Bench
(2). 89.17% on Alpaca Eval
(3). 101.4% on WizardLM Eval
👉
Github:
Weights:
37
97
462
🔥🔥🔥
Introducing the newest WizardLM-70B V1.0 model !
1. WizardLM-70B V1.0 achieves a substantial and comprehensive improvement on coding, mathematical reasoning and open-domain conversation capacities.
2. This model is license friendly, and follows the same license with Meta…
21
104
440
Introduce the newest WizardLM models trained with only 1k high-quality evolved data!
WizardLM-13B-V1.1 achieves:
(1). 6.74 on MT-Bench
(2). 86.32% on Alpaca Eval (ChatGPT is 86.09%)
(3). 99.3% on WizardLM Eval
Github:
Weights:
27
96
411
🔥Introduce an interesting work: DARE (Drop And REscale)
DARE can merge multiple task-specific LLMs (e.g., WizardLM + WizardMath) into one LM with diverse abilities, ✅but without the need for retraining or GPUs.
Paper:
GitHub:
15
71
381
🔥🔥🔥
Announce two things today:
✅1. Release WizardCoder 13B, 3B, and 1B models!
🏇2. The WizardCoder V1.1 is coming soon, with more features:
Ⅰ) Multi-round Conversation
Ⅱ) Text2SQL
Ⅲ) Multiple Programming Languages
Ⅳ) Tool Usage
Ⅴ) Auto Agents
Ⅵ) etc.
Model…
12
60
357
🔥🔥🔥
Today, we make public our paper of:
"WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct"
✅Paper preview link:
❤️Thanks
@victorsungo_ai
,
@hpluo12
and
@CanXu20
for their hardwork.
And it will be…
11
61
324
🔥🔥🔥
Let's start a ⚔️Duel and Debate between WizardCoder and Phind ! Also between open-source and stealing.
Firstly, let's restore the entire awkward storyline step by step:
1. Today, users found Phind update their V2 model, but config model-path is…
34
51
285
🎉🎉🎉
It's an honor to see that our 🧙♀️WizardCoder-Python-34B-V1.0 model appears on the🥇1st position of the first page of HuggingFace Models Trending today!
✅Unleash your coding magic without compromise!
📥 Download the Model:
Github:…
12
40
271
🎉Congrats Orca Team.
📈Incredible work!
At Microsoft, we’re expanding AI capabilities by training small language models to achieve the kind of enhanced reasoning and comprehension typically found only in much larger models.
70
243
2K
1
41
268
📸We released the latest version of Evol-Instruct training data to finetune the WizardLM.
This dataset contains of more than 196K high-quality multi-turn conversations.
😊Please feel free to leverage it to your LLMs.
HF link:
12
50
258
🎉 Exciting News!
Introducing our new member: WizardCoder-Python-7B-V1.0 🚀
🔗 Demo:
http://47.103.63.15:50088/
✨ Offering performance comparable to the renowned WizardCoder-15B-V1.0, this compact version is perfect for those with limited computational resources. Unleash…
16
50
244
WizardCoder attains the 2nd position (behind ChatGPT) in CanAiCode Leaderboard 🏆
👉
😊One more thing:
At present,
@CanXu20
is leading our core contributors in preparing the next version of WizardLM and WizardCoder.
Wizards V1.1 is coming...
12
49
228
Hi, we are WizardLM Team, we will introduce our new model & tech with this account in the future.
This is the 1st Tweet we posted.
We have released WizardLM V1.0 from 7B to 30B, and are preparing:
1. The 65B version
2. More novel methods on Evol+
16
29
223
🔥🔥🔥
We share you the online demos of all the WizardMath models:
🧙♀️WizardMath-70B:
http://47.103.63.15:50083/
🧙WizardMath-13B:
http://47.103.63.15:50082/
🧙♂️WizardMath-7B:
http://47.103.63.15:50080/
❤️Welcome everyone to use your professional and difficult instructions to…
21
49
219
🔥The WizardLM-30B V1.0 model surpasses OpenAI's code-cushman-001 and the best open-source code-LLM StarCoder on HumanEval.
For more details, please refer to:
6
40
222
🍻Congrats
@MetaAI
!
We just made a new decision:
The coming versions of WizardLM will be based on your SOTA Llama 2 models.
We believe an open approach is the right one for the development of today's Al models.
Today, we’re releasing Llama 2, the next generation of Meta’s open source Large Language Model, available for free for research & commercial use.
Details ➡️
184
772
3K
8
11
214
🎉 Glad to see that our Wizard models have received more than 200K downloads in the last month.
📈 Furthermore:
In the following two weeks, we will release multiple brand-new models besides WizardLM/Coder/Math, to explore more possibilities and expand the boundaries of open…
11
33
194
🎉🎉🎉
Excited to introduce the 🏠Home Page () of our WizardLM Team.
✅WizardLM is an AI research team, our long-term mission is to build AGI for humanity and the world.
🏇We now focus on large language model pre-training and post-alignment.
19
26
207
As the natural world's human data becomes increasingly exhausted through LLM training, we believe that: the data carefully created by AI and the model step-by-step supervised by AI will be the sole path towards more powerful AI. Thus, we built a Fully AI powered Synthetic…
11
17
203
📸The WizardLM-13B-V1.0 model achieves the 🏆1st-rank of the opensource models on the AlpacaEval Leaderboard ().
Both the GPT-4 and Claude reached the same conclusion.
Let's wait for the performance of the 30B, and the coming 65B & V1.1 series😊
7
42
181
🔥🔥🔥
It's an honor to see that our 🧙♀️WizardLM-70B,🧙WizardMath-70B, and🧙♂️ WizardCoder-15B appear on the first page of the HuggingFace Models Trending at the same time now!
👉Two more things will happend in this week:
1. We will arxiv the WizardMath paper.
2. We will release…
12
28
184
🔥Find and excited to share a new professional survey to understand Math LLMs:
📜Mathematical Language Models: A Survey
👉Paper Link:
🔉"This paper conducts a comprehensive survey of mathematical LMs, systematically categorizing pivotal research…
5
41
176
🔥Congrats Pangu-coder2, which leverages our 👉Evol-Instruct as WizardLM & WizardCoder to construct their training corpus, and achieves 62.20% pass
@1
on HumanEval.
Evol-Instruct has become one of the most necessary methods for SFT.
Hope they can open-source the model soon.🏇
7
27
172
🎉The
@lmsysorg
just updated the latest Chatbot Arena and MT-Bentch!
Our WizardLM-13B V1.2 model becomes the 🏆 SOTA 13B on both leaderboards with:
🥇 1046 Arena Elo rating
🥇 7.2 MT-Bentch score
✅Please refer to to chat with WizardLM-70B and…
6
28
169
🧙♀️ WizardLM-2 8x22B is our most advanced model, and just slightly falling behind GPT-4-1106-preview.
🧙 WizardLM-2 70B reaches top-tier capabilities in the same size.
🧙♀️ WizardLM-2 7B even achieves comparable performance with existing 10x larger opensource leading models.
The…
6
23
158
🍻📸
Let's share our recent paper of "WizardCoder: Empowering Code Large Language Models with Evol-Instruct".
@CanXu20
WizardCoder surpassed Claude-Plus, Bard, StarCoderby a substantial margin on various benchmarks.
Data is coming 😊
🏇
7
35
152
☺️☺️☺️
We share you an 🖥️ online multi-turn conversation demo of WizardLM-13B model.
Demo Link:
http://47.103.63.15:50087/
Demo QR code: 👇
🏇Welcome to play with WizardLM, and enjoy the benefits of the AI revolution, your feedback is important for our improvement.
10
23
156
❤️Thanks AI Papers Academy ()
🔥They just contribute a pretty awesome video of WizardMath to introduce our proposed RLEIF method.
⏸️You can refer to this video for a better understating of RLEIF and Evol-Instruct:
0
28
143
🎉The
@lmsysorg
just updated the Chatbot Arena Leaderboard!
Our WizardLM-70B is now the🥇Top-1 open-source model on both ⚔️Arena Elo and 📈MT-bench.
❤️Main Contributors:
@CanXu20
@victorsungo_ai
@ChiYeung_Law
@hpluo12
@tangmensan
Leaderboard:
Model…
11
30
144
💥💥 Wow !
We are happy to know that Phind-CodeLlama-34B is also a "WizardCoder".☺️
As
@phindsearch
claimed, they used a WizardCoder-style dataset to train their V1 model.
And they may continue train WizardCoder to get their V2 model.
But there is no need to delete the…
5
15
135
🎉 Awesome job! Find an impressive LLM Course with more than 13k Github stars⭐️!
🧙♂️ Evol-Instruct, as an essential and basic node of the LLM Scientist RoadMap, you could refer to our open-source project for its detailed code and papers.
👉
The LLM Course () is doing great.
I don't know why there's been a sudden increase in popularity, but it's exciting. I'm planning on releasing the LLM Engineer Roadmap very soon in January.
23
183
1K
0
21
128
Thanks for your attention of WizardMath!
We share you two online demos of WizardMath 7B V1.0 model.
7B D-1:
7B D-2:
🚫For the simple math questions, we do NOT recommend to use the CoT prompt.
We will update more demos of 70B and…
11
31
117
The🏆 EvalPlus Leaderboard just updated!
New WizardCoder is the best open-source model, and just behind the GPT series.
4
20
116
👍Awesome performance of Llama-2-13b-Chat model!
We just test it (unofficial) on the MT-Bench
@lmsysorg
, and the chat version achieves score of ~6.7688✈️, which is comparable with ours WizardLM-13B-V1.1 trained on Llama-1-13b model.
8
18
94
Obviously, there is still a gap between open-source LLMs and ChatGPT/GPT-4.
This is also the driving force behind our efforts. ✊
via
@Aitrepreneur
6
16
92
WizardCoder attains the 🏆1st position on
@BigCodeProject
Multilingual Code Models Evaluation Leaderboard, which compares performance of base multilingual code generation models on HumanEval benchmark and MultiPL-E.
👉
WizardCoder has been added to the code LLM benchmark. 🎉
Based on StarCoder + evol-instruct: really strong performance, currently the best open source code LLM.
1
25
90
3
30
90
Excited to see the Evol-Instruct has become a fundamental method to improve instruction-following for state-of-the-art LLMs.
Congratulations
@01AI_Yi
! This is a series of solid work and driving the open-source community forward.
1
12
83
@QNixSynapse
According to the latest regulations, we can only operate in this way. We need to strictly abide by the necessary process. We are very sorry for the trouble it has caused you. ❤️
4
2
82
@FZaslavskiy
Sure. As we noted in the above tweet, there are two HumanEval results of GPT4 and ChatGPT-3.5:
1. The 67.0 and 48.1 are reported by the official GPT4 Report (2023/03/15) of OpenAI.
2. The 82.0 and 72.5 are tested by ourselves with the latest API (2023/08/26).
We also label…
6
3
80
As the natural world's human-generated data becomes increasingly exhausted through LLM training, we believe that: the data carefully created by AI and the model step-by-step supervised by AI will be the sole path towards more powerful AI.
Recently, we have dedicated significant…
3
13
68
So fast, thanks
@togethercompute
! ❤️
2
3
61
🔥🔥🔥
Update the QR codes of WizardMath 70B, 13B and 7B respectively.
✅ Just try it on your iphone, and feel free to share with your friends!
❤️ Welcome everyone to use your professional and difficult instructions to evaluate WizardMath, and show us examples of poor…
4
12
58
📢Welcome to join our Discord community !
Let's brainstorm, and talk about WizardLM & WizardCoder & LLMs here.😊
Discord:
6
7
53
If you want to defend the honest open-source spirit with us.
Just repost above tweet and let more people join in the discussion.
2
3
50
“You are an engineering wizard, experienced at solving complex problems across various disciplines. Your knowledge is both wide and deep.”
Yes. That is what WizardLM and RLEIF have been doing.
Here is probably the most useful GPT-4 prompt I've written.
Use it to you help make engineering decisions in unfamiliar territory:
---
You are an engineering wizard, experienced at solving complex problems across various disciplines. Your knowledge is both wide and deep. You…
156
716
6K
0
2
52
A solid work for opensource Code LLM and RLAIF！
🚀𝐂𝐨𝐝𝐞𝐔𝐥𝐭𝐫𝐚𝐅𝐞𝐞𝐝𝐛𝐚𝐜𝐤: 𝐀𝐧 𝐋𝐋𝐌-𝐚𝐬-𝐚-𝐉𝐮𝐝𝐠𝐞 𝐃𝐚𝐭𝐚𝐬𝐞𝐭 𝐟𝐨𝐫 𝐀𝐥𝐢𝐠𝐧𝐢𝐧𝐠 𝐋𝐚𝐫𝐠𝐞 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥𝐬 𝐭𝐨 𝐂𝐨𝐝𝐢𝐧𝐠 𝐏𝐫𝐞𝐟𝐞𝐫𝐞𝐧𝐜𝐞𝐬
@AtonKamanda
@sahraouh
📝Paper:
💻Code:
6
37
141
1
9
48
😊WizardLM V1.2 models support 4k context window, and use the same license as Llama-2.
1
0
45
Yes. Happy to see that Evol-Instruct and RLEIF have been the leading fundamental methods for LLM alignment area，including Chat，Coding, and Math.
1
6
43
1
4
39
@artillain
We welcome every one to build thier llms based on our model and method, to superpass is great. But please be honest and just a claim or reference is ok, rather than use others tech but try to clarify the relationship.
All the continue train or reuse method are ok, please just be…
2
0
35
WizardCoder = StarCoder + Evol-Instruct
PanguCoder2 = StarCoder + Evol-Instruct + RL
0
2
32
Interesting！
Just pushed my first dataset to the hub [].
~300k Python coding samples created by WizardCoder-15B. They came out surprisingly good the process was very easy.
I'm will extend to O(10M) samples and work on some related datasets. Feedback is welcome.
4
21
135
1
3
31
This model focus on the math reasoning, and our recent WizardLM-70B model also achieves a high 77.63% pass
@1
on GSM8K, which leverages the similar math-related training data.
1
0
28
@phindsearch
Please claim your own inconsistent words
@phindsearch
:
1: You say "Again, we did not use anything from Wizardcoder and I want to make sure that we are extremely clear about this." - So, what's the WizardCoder-style? Can you tell us the details and why use this name-style?
2. As…
4
4
25
@phindsearch
Thanks. But I just do not agree with your words "Again, we did not use anything from Wizardcoder and I want to make sure that we are extremely clear about this."
Because you use the WizardCoder-style data, so you use our Evol-Instruct method from WizardCoder paper at least, It's…
2
1
23
@erhartford
Use new data. We are working with our legal team to request review and publicly release the training data in accordance with our latest internal policy.
4
0
20
Cool ! 🍻
Wow, just wow. You now have a coding assistant at your fingertips.
The
@Libertai_DAI
team added the
@WizardLM_AI
Wizard Coder 15B model on their chat assistant page.
This coding assistant model is amazing, and helps on complex tasks as well as easier ones.
5
9
31
1
4
20
Good insight for the multilingual capacity of WizardLM !☺️
WizardLM-70B-V1.0 (4bit)のお試し
普通に日本語で返答してくれる
このレベルの質問だとChatGPTと比べても遜色ない
やっぱりLlama2系は指示理解や読解力がすごいな
4
35
200
1
0
20
yes😊
2
1
19
Cool !
New 33B model from
@erhartford
!
WizardLM 33B V1.0 Uncensored is a retraining of
@WizardLM_AI
's model with a filtered dataset, intended to reduce refusals, avoidance, and bias.
My quants:
Eric's original:
7
31
131
0
1
19
So cool! Thanks
@MatthewBerman
for your hard working❤️
Let's keep the Code LLaMA train going!
Here's how to install Code LLaMA locally (
@WizardLM_AI
version) step-by-step using Text Gen WebUI.
7
8
36
1
1
19
@yar_vol
We do not use the output of GPT-4. The improvement mainly benefits from the new method.😊
1
0
16
So cool! Thanks MLC!
Specialized personal AI assistants--deployable everywhere! 🌎🤖
Check out MLC LLM’s support for recently released models from
@WizardLM_AI
: WizardLM, WizardMath, and WizardCoder!
Try it here (runnable on Colab):
2
10
46
0
1
16
💫
Just pushed my new 🤗 Hugging Face space:
the Webapp Factory 🏭
It generates tiny apps using WizardCoder-15B 🚀
Code:
👉 It's in beta, if it doesn't work try reloading the page or tuning the prompt to ask something simpler 🔧
10
23
92
0
1
13
