AlpinDale
|
41beab5dc1
add exllamav2 tensor paralell, fused MoE for GPTQ/AWQ
|
преди 10 месеца |
AlpinDale
|
8fa608aeb7
feat: replace Ray with NCCL for control plane comms (#221)
|
преди 1 година |
AlpinDale
|
801eda0b7a
feat: support GPTQ 2, 3, and 8bit quants (#181)
|
преди 1 година |
AlpinDale
|
b9b295d74e
chore: backlogs 1 (#191)
|
преди 1 година |
AlpinDale
|
62b2c4119d
feat: re-write GPTQ and refactor exllama kernels (#152)
|
преди 1 година |
AlpinDale
|
2b1ba581f9
feat: re-implement GPTQ (#141)
|
преди 1 година |
AlpinDale
|
887e03669a
feat: add exllamav2 for GPTQ (#99)
|
преди 1 година |