i386

Its an entire misunderstanding of how the layout system works.

Most people working at the top of these large organizations don’t know how to make anything net new anymore. They pushed all the people that do out when padding their nests.

exo is Mac only.

I’d love a bug report - we can get it working for you!

Our skippy library is a patch queue on top of llama that allows us to access internal information, such as activations, and filter tensors on model load.

To be honest, both are very tough problems we don't have a good answer for yet. If that is something that concerns you, look into building a private mesh with trusted peers.

Each stage has its own KV for the layers it hosts. You are on the money there, when one stage is waiting it's free for more parallelism. I am planning on exploiting this for more token verification through ngram spec decoding.

It can dynamically route. If a machine drops out of split, the topology is recalculated and the request is automatically retried.

Thank you! AMD is a weak spot in our testing right now. If you’re willing to contribute or let us borrow some compute time, drop in on the Discord.

The lab features two Mac Studios: an Apple M3 Ultra (32 CPU cores, 80 GPU cores, 256 GB unified memory) and an Apple M1 Ultra (20 CPU cores, 48 GPU cores, 128 GB unified memory), both connected via 1Gbit Ethernet.

We use a customized Q2 quantization that preserves sensitive tensors at Q8.

To reduce compute time per layer, we are developing a custom GLM DSA Metal graph.

While we are not yet approaching MTP, we plan to port our existing MTP implementations from versions 4.7 and 5.1 to 5.2.

Since GLM's MTP acceptance rate is very high for a single predicted token, we are exploring token prediction techniques to widen the predicted tokens and utilize parallelism for verification.

That isn’t true. llama RPC is incredibly slow but staged splits in skippy are orders of magnitude faster.

This was done on my home lab simulating 5ms latency and jitter between machines. Splits work quite well if you your nodes are over WAN at metro latency’s but not super fast on global WAN.

The idea is that you could take several machines without dedicated RDMA or NVLINK fabric and use them to serve a large model on hardware you own then share it with others.

I’m currently working on GLM 5.2 on my lab environment with around 10 tok/s on the same split.

I’m one of the contributors to Mesh LLM and happy to answer any questions. I authored the skippy engine that allows you to split large models across nodes.

PMs at Microsoft have incredibly bad taste