huggingface/text-generation-inference

Pull Requests Commits

Fixup constructor arguments

danieldk committed 1 year ago

959b9dc2

Sync allocator interfaces

danieldk committed 1 year ago

27ef5aa0

Last accessed fixes

danieldk committed 1 year ago

48b21eab

Proper support for two allocations with overlapping prefixes

danieldk committed 1 year ago

dd2d6cfe

Add another problematic case

danieldk committed 1 year ago

d4ce5389

danieldk committed 1 year ago

0e6ff129

First step towards cleaning up

danieldk committed 1 year ago

7c046c91

Renaming, window size

danieldk committed 1 year ago

05611f6b

Traitify the current allocator in preparation for swappable alloc

danieldk committed 1 year ago

083806aa

Better leaf tracking

danieldk committed 1 year ago

3b4754cd

Avoid continuous sorting during reclamation

danieldk committed 1 year ago

1a461234

Shake out some issues, add correct removal order test

danieldk committed 1 year ago

c352a3e2

danieldk committed 1 year ago

6d0094e5

Walk up to predecessors

danieldk committed 1 year ago

3b6bef40

Basic test passes

danieldk committed 1 year ago

9da64a7b

danieldk committed 1 year ago

dbb82e27

Use symmetric quantization in the `quantize` subcommand (#2120)

danieldk committed 1 year ago

Verified dbb23fbf

[fix] Modifying base in yarn embedding (#2212)

SeongBeomLEE committed 1 year ago

Verified c46eaf70

fix: append DONE message to chat stream (#2221)

drbh committed 1 year ago

Verified d789de32

Add support for FP8 on compute capability >=8.0, <8.9 (#2213)

danieldk committed 1 year ago

Verified cb150eb2

Move quantized weight handling out of the `Weights` class (#2194)

danieldk committed 1 year ago

Verified 8511669c

Updating the self check (#2209)

Narsil committed 1 year ago

Verified 4c976fb4

Fixed README ToC (#2196)

vinkamath committed 1 year ago

Verified f5ba9bfd

Adding sanity check to openapi docs.

Narsil committed 1 year ago

Verified fe710af2

Fix buildx cache + change runner type (#2176)

glegendre01 committed 1 year ago

Verified 5e2a3058

Fix nccl regression on PyTorch 2.3 upgrade (#2099)

fxmarty committed 1 year ago

Verified 4c50b6d0

feat: use model name as adapter id in chat endpoints (#2128)

drbh committed 1 year ago

Verified 87ebb647

update to metrics 0.23.0 or could work with metrics-exporter-promethe… (#2190)

sywangyi committed 1 year ago

Verified 58effe78

fix: python deserialization (#2178)

jaluma committed 1 year ago

Verified 16d9e505

add doc for intel gpus (#2181)

sywangyi committed 1 year ago

Verified 07e240ca

Older