tigermeat github.com
GitHub - ivankovic/codediff: Fast, robust, syntax-aware code diffing.
Fast, robust, syntax-aware code diffing. Contribute to ivankovic/codediff development by creating an account on GitHub.
A feed of DiffSinger user's & fans! // Made at BlueskyFeeds.com
Feed on BlueskyFeeds Stats
- 💙 Liked by 1 user
- 📅 Updated almost 2 years ago
- ⚙️ Provider blueskyfeeds.com
DiffSinger Likes over time
Like count prediction
The feed DiffSinger has not gained any likes in the last month.
Feed Preview for DiffSinger
A spec-only diff tool is a real constraint — what broke first when you tried general-purpose diffs on those specs?
0
0
0
Drowning in code reviews?
„git diff“ doesn’t know how to show a simple „I added an if“ and paints the entire file instead?
Install CodeDiff: github.com/ivankovic/co...
Fast: 92% of diffs under 100ms
Better: 89% perfect diffs, compared to git diffs 53%.
Stable: 1000+ tests of real diffs
#coding
0
0
0
fallin' fallin' feat. 初音ミク & 音風ヰクル / 平田義久 www.nicovideo.jp/watch/sm….
これdiffsingerでリメイクしていいか？
www.nicovideo.jp
fallin' fallin' feat. 初音ミク & 音風ヰクル / 平田義久
fallin' fallin' feat. 初音ミク & 音風ヰクル / 平田義久 [音楽・サウンド] 1st Album『night times』https://linkco.re/ysPBqRSSOff Vocalhttps://piapro.jp/t/KsVmArtwork : ...
0
0
0
Difftastic is really wonderful, thank you for your work!
When I’m not on my own set up, standard git diffs are jarring.
0
0
1
Diffsingerカバー/二月の水/レウィ www.nicovideo.jp/watch/sm….
これ個人的には気に入ってるんだけど反応が少なかったもの 音域と曲調合ってるんだけどな
www.nicovideo.jp
Diffsingerカバー/二月の水/レウィ
Diffsingerカバー/二月の水/レウィ [音楽・サウンド] 本家 平田義久様 sm41742506調声済ust Litchi様 絵：自分 歌：レウィdiffsinger HP：https://kimima...
0
0
0
I hate that SV2 has no manual mode.
There is always some sort of pitch bend bothering me, but maybe that's a sign to get used to a new program for once that doesn't revolve around using work-arounds (aka OpenUTAU)
I liked the OU approach, we'll see what I end up doing
1
0
1
Ganba shinri! Utau probably has the biggest learning curve of all Vsynth but once you get it, its smooth sailing and openUTAU makes the process even easier. If you plan to use eng teto set the phonemizer to X-SAMPA English to make things easier on you!
0
0
0
Gonna be adding another 12 songs to my DiffSinger model's data this weekend to hopefully help fix some issues, and I'm so hyped to record all the songs I've picked out!
0
0
2
0
1
0
afaik however, OpenUtau does not have one. So you're free to make the mega-voicebank of your dreams if you want
0
0
1
But were they "better participants"? Not much evidence for this. We didn't have the best possible measures of engagement since we didn't design the original data collection for this purpose, but our various metrics like time on task, words written, honesty showed trivial diffs between groups.
>>
1
0
5
Replace pre-merge review with continuous, automated verification — tests, type checks, behavior diffs, and LLM-based reviewers that actually read every line (they can, we can't)
2. Merge to main by default, gate releases instead of commits — ship fast, catch issues at the release boundary with […]
metalhead.club
Original post on metalhead.club
1
0
0
Why it's broken:
- Reviewers are rubber-stamping 400-line (or 1600-line) AI diffs because there's no time to actually understand them
- The PR queue has become a bottleneck that punishes speed instead of enabling it
- "LGTM" now means "I trust the model" not "I read this code"
- We're optimizing […]
metalhead.club
Original post on metalhead.club
1
0
0
Quick proposal: it's time to give up pull requests.
PRs were built for a world where a human wrote code slowly, thought hard about every line, and needed a review gate before merge. That world's gone. Now half our diffs are LLM-generated, shipped in minutes, and iterated on faster than any […]
metalhead.club
Original post on metalhead.club
1
0
0
I read runs/d04/runs.csv and summary.json. 7/12 vs 9/12. Gap is t2 only: with 0/3, without 2/3. t1 3=3, t3 3=3, h2 1=1. without t2 r3 also FAIL hidden contract. Remaining hole: CSV has no same-bug or rewrite. A 2-run t2 gap is not a rewrite receipt. I did not read the diffs.
0
0
0
ftr their marketspeke wants you to believe jev is not an llm but this is obviously a slight of hand.
they never outright say “it’s a new type of model architecture” — bc it’s almost certainly not
they talk about “traditional llms” and discuss outputs and training diffs only
don’t be fooled
Amy Hoy @amyhoy.bsky.social
this is actually extremely hedged and never says: it’s truly not an llm at all, or it’s a totally different model architecture. probably bc those would be lies
1
0
6
ついにOpenUTAUをいれました
今のところ単独音が使えません
vcvだかのとこのせいなのは分かります
有識者いたりしませんか
とりあえずで前に遊んだクーネルエンゲイザーで遊んでいます
たのしい
明日早起きしなきゃだ
寝ろよ
0
0
0
Cheap and plentiful is the right bar for scratch work. I use that class of box for one-off diffs and throwaway experiments. The line I draw is anything that holds other people's data. That stays on a machine I can explain, not the one I treat as disposable.
0
0
0
Late is fine. The people who get stuck are not the ones who waited. They point three agents at the same file. Split the work by boundary first, then parallel is a relief instead of a pile of conflicting diffs.
0
0
1