huggingface/text-generation-inference

Pull Requests Commits

Upgrade intermediary layer for nvidia too. (#1557)

Narsil committed 1 year ago

Verified 7671a419

Fixing glibc version in the runtime. (#1556)

Narsil committed 1 year ago

Verified 6f68bb14

feat: add deserialize_with that handles strings or objects with content (#1550)

drbh committed 1 year ago

Verified 246ad39d

feat: experimental support for cuda graphs (#1428)

OlivierDehaene committed 1 year ago

Verified 0d794af6

feat(router): add max_batch_size (#1542)

OlivierDehaene committed 1 year ago

Verified 53214633

ROCm AWQ support (#1514)

IlyasMoutawwakil committed 1 year ago

Verified a4e58016

chore: bump ci rust version (#1543)

drbh committed 1 year ago

Verified c5ef81be

feat(server): add frequency penalty (#1541)

OlivierDehaene committed 1 year ago

Verified 09b7c26b

Update to peft 0.8.2 (#1537)

Stillerman committed 1 year ago

Verified 39af000c

Impl simple mamba model (#1480)

drbh committed 1 year ago

Verified bd405e03

feat: use existing add_generation_prompt variable from config in temp… (#1533)

drbh committed 1 year ago

Verified 17345402

feat: add ie update to message docs (#1523)

drbh committed 1 year ago

Verified 0da00be5

[docs] Fix link to Install CLI (#1526)

pcuenca committed 1 year ago

Verified 3ab578b4

Updating tokenizers. (#1517)

Narsil committed 1 year ago

Verified 0e97af45

fix: tokenizer config should use local model path when possible (#1518)

drbh committed 1 year ago

Verified ee1cf51c

Revert "Modify default for max_new_tokens in python client (#1336)"

Narsil committed 1 year ago

1e03b61b

Hotfix the / health - route. (#1515)

Narsil committed 1 year ago

Verified 9ad7b6a1

Freshen up the README.

Narsil committed 1 year ago

94d243b3

GPTNeoX: Use static rotary embedding (#1498)

dwyatte committed 1 year ago

Verified 13c62be4

fix: improve messages api docs content and formatting (#1506)

drbh committed 1 year ago

Verified 2ae36a97

feat: eetq gemv optimization when batch_size <= 4 (#1502)

dtlzhuangz committed 1 year ago

Verified 0595bf3e

Modify default for max_new_tokens in python client (#1336)

freitng committed 1 year ago

Verified 2d56f106

Create the compute type at launch time (if not provided in the env). (#1505)

Narsil committed 1 year ago

Verified a9ea6068

Sending compute type from the environment instead of hardcoded string (#1504)

Narsil committed 1 year ago

Verified 0424dabb

Fixing top_n_tokens. (#1497)

Narsil committed 1 year ago

Verified 069895b9

OlivierDehaene committed 1 year ago

Verified c2d4a3b5

feat: add tokenizer-config-path to launcher args (#1495)

drbh committed 1 year ago

Verified d9758851

GPTQ support on ROCm (#1489)

fxmarty committed 1 year ago

Verified 650fea18

Update the docs to include newer models. (#1492)

Narsil committed 1 year ago

Verified ebecc061

fix: launcher doc typos (#1462)

thelinuxkid committed 1 year ago

Verified 50a20a83

Newer Older