thaw-vllm
The fork primitive for LLM inference. Snapshot a running session — weights + KV cache + scheduler state — and hydrate it into N divergent children that skip prefill. For RL rollouts, parallel coding agents, agent branching. Supports vLLM and SGLang.
Version0.5.1
Downloads
7.28k
LicenseApache-2.0
AuthorNils Matteson
Downloads
Weekly, last 90d.
Includes CI traffic.
VersionsTotal0.*
Range
View
Granularity
Group by
CI traffic
Stack: OffCI: Included2 / 16 series
Selected total14.55k
7.3kAll-time
1.4kLast 30 days
20Last 24 h
<0.01/sPer second
Sponsored
Sponsorships keep pepy free to read
Version distribution
Share of downloads by released version. Computed over the last quarter.
- 0110.2%
0.3.0
740 downloadsDownloads74010.2% - 028.6%
0.1.0
627 downloadsDownloads6278.6% - 038.3%
0.2.1
606 downloadsDownloads6068.3% - 048.3%
0.5.1
602 downloadsDownloads6028.3% - 056.6%
0.1.4
480 downloadsDownloads4806.6% - 066.4%
0.2.0
468 downloadsDownloads4686.4% - 076.3%
0.1.2
455 downloadsDownloads4556.3% - 086.2%
0.1.3
451 downloadsDownloads4516.2% - 0939.1%
Other
2.8k downloadsDownloads2.8k39.1%
Guess the next day
Thirteen recent days of thaw-vllm downloads. Drag the green handle on the right to guess where day fourteen lands.