woctordho

7 ms·

61.
by woctordho 7mo ago
RCE is exactly the feature of coding agents. I'm happy with it that I don't need to launch OpenCode with --dangerously-skip every time.
62.
by woctordho 7mo ago
The real barrier is not just the language keywords but lots of documentations and discussions in English. I'm not sure whether there is a solution to this.
63.
by woctordho 7mo ago
Then don't get sorrow killing it. Living things are not so special.
64.
by woctordho 8mo ago
This time even Unsloth could not provide bitsandbytes 4-bit models. bitsandbytes does not support new models with MoE and linear attentions, and it's much less flexible than GGUF. Nowadays I think it's better to train lora over GG
65.
by woctordho 8mo ago
In one word, porn. Qwen filtered out a lot of porn during data curation, and a finetuned model can perform much better than context engineering. Abliteration can only remove censorship, not add something non-existent in the training data. T
66.
by woctordho 8mo ago
It would be good if we can use formal verification to see to which extent the quantization will overflow in intermediate results. There are some widely-known annoying bugs that SageAttention (int8 quantized attention) works on some models b
67.
by woctordho 8mo ago
Here is a recent example that lack of archived discussion causes problem: https://github.com/triton-lang/triton/issues/9539
68.
by woctordho 8mo ago
There are certainly things can't be done via CLI, or more suitable for a persistent daemon with various RPC rather than a CLI. But most things are simpler than that, and MCP is overcomplicating it. MCP does not make things more organiz
69.
by woctordho 8mo ago
If you archive the session, you automatically archive all Google search history (queries and outputs) that the AI did, and it's usually relevant to the project.
70.
by woctordho 8mo ago
You can export it with DataClaw. By default it outputs jsonl and publishes to HuggingFace, but you can also do analysis locally with it.
71.
by woctordho 8mo ago
Ideally, yes. Although Slack is a vendor lock-in and we need a better platform to archive the sessions.
72.
by woctordho 8mo ago
That's exactly one of the reasons I've been archiving the sessions using DataClaw. The sessions can contain more useful information than the comments for humans. [0] https://github.com/peteromallet/dataclaw
73.
by woctordho 8mo ago
Interface for non-devs has evolved. We already have OpenClaw and things like Claude Cowork.
74.
by woctordho 8mo ago
You can just write a README.md and put it along with the CLI
75.
by woctordho 8mo ago
You may check Prime Intellect's prime-diloco and Nous Research's DisTrO
76.
DataClaw: Publish your Claude Code chats to HuggingFace with a single command(github.com)
2 points by woctordho 8mo ago|2 comments
77.
by woctordho 8mo ago
Not my project, but I had a similar idea during the recent discussions about distillation. I'm glad more people are doing this.
78.
by woctordho 8mo ago
Law can only take down people, but not a decentralized autonomous organization (DAO).
79.
by woctordho 8mo ago
"We work by three virtues: rage, paranoia, and kleptomania."
80.
by woctordho 8mo ago
If there was no AA, there would still be another random guy who assembles such datasets and distributes them before LLM companies.
81.
by woctordho 8mo ago
CSAM is not something to scare people away. In P2P networks like Perfect Dark there are TBs of CSAM sitting in everyone's disks and we just get along with them.
82.
by woctordho 8mo ago
If it helps: I'm actively considering how to make LLMs more helpful in GPU programming. Recently I converted the RDNA 3.5 ISA manual from PDF to Markdown for LLM retrieval (because I have a Strix Halo). Notably, I converted every bitfi
83.
by woctordho 8mo ago
I'd say it's pretty good on supported GPUs (gfx 9 and 12). Things like hipcc and Triton all work on top of it. But on gfx 11 there is still a lot to optimize.
84.
by woctordho 8mo ago
Major players in China don't play like that. MooreThreads, Lisuan, and many other smaller companies all have their own porting kits, which are basically copied from HIP. They just port every piece of software and it just works. If you
85.
by woctordho 8mo ago
I'm also wondering this. The compiler itself is written in C99, but looking from the tests, it can parse some C++ features such as templates.
86.
by woctordho 8mo ago
Yes. It outputs a hsaco binary that just runs on the GPU (as long as you have the driver). No ROCm needed.
87.
by woctordho 8mo ago
They've been doing it all the time and it's called HIP. Nowadays it works pretty well on a few supported GPUs (CDNA 3 and RDNA 4).
88.
by woctordho 9mo ago
What a pity that although webtorrent support is already merged in the libtorrent master branch years ago, it's not merged into the stable branch yet, therefore not working out of the box in clients like qBittorrent.
89.
by woctordho 10mo ago
It takes time and effort to receive money, especially from consumers worldwide. Most hobbyists would not going to deal with all the complexity in it.
90.
by woctordho 10mo ago
There is a pretty interesting open source music AI named ACE-Step. Currently its quality is at about the Stable Diffusion 1.0 level, and they'll release a new version soon (hopefully in January).

More ›