Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
Der_Einzige
23 days ago
|
parent
|
context
|
favorite
| on:
GLM-4.7-Flash
Speculative decoding isn’t needed at all, right? Why include the final bits about it?
latchkey
23 days ago
|
next
[–]
https://docs.vllm.ai/projects/recipes/en/latest/GLM/GLM.html...
foobar10000
23 days ago
|
prev
[–]
GLM 4.7 supports it - and in my experience for Claude code a 80 plus hit rate in speculative is reasonable. So it is a significant speed up.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
