Creating training data for software engineering agents is difficult. Until now.
Introducing SWE-smith: Generate 100s to 1000s of task instances for any GitHub repository.
We've generated 50k+ task instances for 128 popular GitHub repositories, then trained our own LM for SWE-agent.
The result? SWE-agent-LM-32B achieve 40% pass@1 on SWE-bench Verified.
Now, we've open-sourced everything, and we're excited to see what you build with it!
Check out the tutorial below to generate 100 task instances for any GitHub repository in 10 minutes.
Abstract page for arXiv paper 2505.03335: Absolute Zero: Reinforced Self-play Reasoning with Zero Data
Another F/A-18 Super Hornet fighter jet from the USS Harry S. Truman aircraft carrier has been lost in the Red Sea, the second jet lost from the carrier in just over a week, four people familiar with the matter told CNN.
Left out of those negotiations were the Memphis City Council; the county Board of Commissioners; and TVA, whose board ultimately had to approve any requests for more than 100 MW of power.
They only learned about the deal in early June, after the deal was done, when Musk announced it with a post on X referencing Memphis’ illustrious music history: “Elvis is in the building.”
That includes the TVA board, which was planning to consider new rates for data centers, including xAI, at its May board meeting. But the board can no longer approve resolutions after President Donald Trump fired two of its members. Consideration of the data center measure has been postponed, TVA spokesperson Scott Brooks said.
apparently this is somehow bad
"Trump is now friends with Carney. Carney is friends with Trump and Canada is screwed."
"I'll be an eight-year president, I'll be a two-term president. I always thought that was very important," Trump told NBC's Meet the Press with Kristen Welker in an interview that aired on Sunday.
Trump, 78, has previously said that he was "not joking" about wanting to serve a third, or even fourth, term as US president.
He later said his statements were meant to troll the "fake news media".
His company, The Trump Organization, has been selling "Trump 2028" hats, fuelling armchair speculation that he may seek to remain in office after his second term ends in January 2029.
"Without DRS being available next year, with very low downforce on the on the straights as we're trying to recharge these batteries would it be a bad thing to have a push to pass?
Endorsements By Political Party (1980-2021)
Progressive Conservative/Conservative: 115 (56 per cent)
Liberal: 41 (20 per cent)
None: 30 (14.5 per cent)
Bloc Québécois: 8 (3.8 per cent)
Mixed: 5 (2.4 per cent)
Canadian Alliance: 4 (1.9 per cent)
NDP: 2 (.9 per cent)
Reform: 1 (.5 per cent)Endorsements By Publication (1980-2021)
Globe and Mail
8 Conservative (1980, 1984, 1988, 1997, 2006, 2008, 2011, 2015)
3 Liberal (1993, 2000 2004)
2 None (2019, 2021)
Toronto Star
12 Liberal (1980, 1984, 1988, 1993, 1997, 2000, 2004, 2006, 2008, 2015, 2019, 2021)
1 NDP (2011)
National Post
7 Conservative (2004, 2006, 2008, 2011, 2015, 2019, 2021)
1 Canadian Alliance (2000)
Toronto Sun
12 Conservative (1980, 1984, 1988, 1993, 1997, 2004, 2006, 2008, 2011, 2015, 2019, 2021)
1 Mixed (2000)
Vancouver Sun
9 Conservative (1980, 1984, 1988, 2006, 2008, 2011, 2015, 2019, 2021)
2 Liberal (1993, 1997)
1 Canadian Alliance (2000)
1 None (2004)
The Province
10 Conservative (1980, 1984, 1988, 2004, 2006, 2008, 2011, 2015, 2019, 2021)
1 None (1993)
1 Liberal (1997)
1 Canadian Alliance (2000)
Hamilton Spectator
5 Liberal (1993, 2000, 2004, 2015, 2019)
4 Conservative (1988, 1997, 2006, 2011)
4 None (1980, 1984, 2008, 2021)
Calgary Herald
11 Conservative (1980, 1984, 1988, 1993, 2004, 2006, 2008, 2011, 2015, 2019, 2021)
1 None (1997)
1 Canadian Alliance (2000)
Edmonton Journal
8 Conservative (1980, 1984, 2006, 2008, 2011, 2015, 2019, 2021)
4 None (1993, 1997, 2000, 2004)
1 Mixed (1988)
Ottawa Citizen
10 Conservative (1980, 1984, 1988, 2004, 2006, 2008, 2011, 2015, 2019, 2021)
2 Liberal (1993, 1997)
1 None (2000)
Montreal Gazette
9 Conservative (1980, 1984, 1988, 2006, 2008, 2011, 2015, 2019, 2021)
4 Liberal (1993, 1997, 2000, 2004)
Waterloo Region Record
6 Liberal (1993, 1997, 2000, 2004, 2015, 2019)
3 Conservative (2006, 2008, 2011)
1 None (2021)
Windsor Star
10 Conservative (1980, 1984, 1988, 2004, 2006, 2008, 2011, 2015, 2019, 2021)
1 Liberal (1993)
1 Reform (1997)
1 None (2000)
La Presse
6 None (1984, 2004, 2008, 2011, 2019, 2021)
5 Liberal (1980, 1993, 1997, 2000, 2015)
2 Conservative (1988, 2006)
StarPhoenix
8 None (1980, 1984, 1988, 1993, 1997, 2004, 2006, 2008)
4 Conservative (2011, 2015, 2019, 2021)
1 Mixed (2000)
Calgary Sun
6 Conservative (2006, 2008, 2011, 2015, 2019, 2021)
Le Devoir
8 Bloc Québécois (1993, 1997, 2004, 2006, 2008, 2011, 2015, 2021)
2 Mixed (2000, 2019)
2 Conservative (1984, 1988)
1 NDP (1980)Endorsements By Owner (1980-2021)
Postmedia: 42
42 Conservative (100%)
CanWest Global: 36
21 Conservative (58%)
8 None (22%)
4 Canadian Alliance (11%)
2 Liberal (6%)
1 Mixed (3%)
Southam: 36
22 Conservative (61%)
7 Liberal (19%)
5 None (14%)
1 Mixed (3%)
1 Reform Party (3%)
TorStar: 26
19 Liberal (73%)
5 Conservative (19%)
1 NDP (4%)
1 None (4%)
Sun Media: 13
12 Conservative (92%)
1 Mixed (8%)
Le Devoir Inc.: 13
8 Bloc Québécois (62%)
2 Conservative (15%)
2 Mixed (15%)
1 NDP (8%)
La Presse Inc.: 13
6 None (47%)
5 Liberal (38%)
2 Conservative (15%)
Thomson: 11
7 Conservative (64%)
2 Liberal (18%)
2 None (18%)
Hollinger: 7
4 Liberal (57%)
2 None (29%)
1 Canadian Alliance (14%)
Sifton: 4
4 None (100%)
NordStar Capital: 3
2 None (67%)
1 Liberal (33%)
Bell Globemedia 3
2 Conservative (67%)
1 Liberal (33%)
trump stays not winning legal cases on the overrreach of his authority. "the first judge to have reached a final decision on the merits" is a Trump appointee. looking forward to finding out if he's a secret left-wing activist judge.
In a week that could only be described as a judicial smackdown, President Trump’s latest batch of executive orders has been met with a resounding “nope” from multiple federal courts. It’s almost as if the Constitution still matters—who knew? First up, Trump’s attempt to require proof of citizenship for voter registration was blocked by a […]
Navigator Research's poll, conducted April 24 to 28 among 1,000 registered voters, found about 1 in 9 (11 percent) of respondents who voted for Trump in 2024 regretted having done so, another 16 percent do not regret voting for him but were disappointed.
In an interview with PsyPost, lead study author Jesús Adrián-Ventura said that he and his team found that right-wing authoritarianism was associated with lower grey matter volume in the dorsomedial prefrontal cortex — a "region involved in understanding others' thoughts and perspectives," as the assistant Zaragoza psychology professor put it.
A grand experiment pitted two rival theories of consciousness against each other. Critics have condemned one of the two theories as pseudoscience, and have questioned whether the new study can deliver a clear verdict on either of them
The Trump administration has dismissed all the scientists and other authors working on the next National Climate Assessment.
The success of Large Language Models (LLMs) has sparked interest in various agentic applications. A key hypothesis is that LLMs, leveraging common sense and Chain-of-Thought (CoT) reasoning, can effectively explore and efficiently solve complex domains. However, LLM agents have been found to suffer from sub-optimal exploration and the knowing-doing gap, the inability to effectively act on knowledge present in the model. In this work, we systematically study why LLMs perform sub-optimally in decision-making scenarios. In particular, we closely examine three prevalent failure modes: greediness, frequency bias, and the knowing-doing gap. We propose mitigation of these shortcomings by fine-tuning via Reinforcement Learning (RL) on self-generated CoT rationales. Our experiments across multi-armed bandits, contextual bandits, and Tic-tac-toe, demonstrate that RL fine-tuning enhances the decision-making abilities of LLMs by increasing exploration and narrowing the knowing-doing gap. Finally, we study both classic exploration mechanisms, such as epsilon-greedy, and LLM-specific approaches, such as self-correction and self-consistency, to enable more effective fine-tuning of LLMs for decision-making.
“We’re taking care of their military. We’re taking care of every aspect of their lives, and we don’t need them to make cars for us,” Trump told Time. “In fact, we don’t want them to make cars for us. We want to make our own cars. We don’t need their lumber. We don’t need their energy. We don’t need anything from Canada. And I say the only way this thing really works is for Canada to become a state.”
U.S. Defense Secretary Pete Hegseth had an internet connection that bypassed the Pentagon’s security protocols set up in his office to use the Signal messaging app on a personal computer, two people familiar with the line told The Associated Press.
While Signal offers more protections than standard text messaging, it’s no guarantee of security. Officials also must ensure their hardware and connections are secure, said Theresa Payton, White House chief information officer under President George W. Bush and now CEO of Fortalice Solutions, a cybersecurity firm.