PygmalionAI aphrodite-engine issues

PygmalionAI / aphrodite-engine

PygmalionAI's large-scale inference engine

https://pygmalion.chat

GNU Affero General Public License v3.0

568 stars 75 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

[Installation]: Upload Aphrodite v0.5.2 On Pypi.org

#451 Abulhanan opened 7 hours ago
0
feat: add batch tokenization endpoint along with option for no token ids

#450 ahme-dev opened 1 day ago
0
Fix minor bugs in outlines and lmfe.

#449 sgsdxzy opened 3 days ago
1
[Installation]: ValueError: 17 is not a valid GGMLQuantizationType

#448 Abulhanan closed 3 days ago
21
[Performance]: Memory Usage Fix for gguf.

#447 Abulhanan closed 4 days ago
3
[Usage]: Please provide the environment variable that closes the KoboldAI Lite page.

#445 online2311 opened 4 days ago
0
fix: restore backwards compatibility with sm_60 (P100 and GP100)

#444 AlpinDale closed 4 days ago
1
Fix out-of-range token crash in OpenAI endpoint

#443 50h100a closed 6 days ago
0
Fix non-batched implementation of logit bias

#442 50h100a closed 6 days ago
0
Tiny fixes

#441 sgsdxzy closed 6 days ago
1
chore: clean up the quants

#440 AlpinDale closed 4 days ago
0
fix: OPTIONS requests in the API

#439 AlpinDale closed 1 week ago
0
feat: ngram prompt lookup decoding

#438 AlpinDale closed 1 week ago
0
Fix/exl2 split

#437 sgsdxzy closed 1 week ago
0
Fix/cohere

#436 sgsdxzy closed 1 week ago
0
[Bug]:

#435 someoneexistsontheinternet opened 1 week ago
0
[Bug]: Unable to use OpenAI API with an auth key via a web browser due to OPTIONS preflight request returning 401.

#434 LostRuins opened 1 week ago
1
Update Kobold Lite Embed

#433 Pyroserenus closed 1 week ago
0
feat: Speculative Decoding using a draft model

#432 AlpinDale closed 1 week ago
0
Abort requests when the connection to /v1/completions is interrupted.

#431 sgsdxzy closed 1 week ago
0
Fix linear bias of qkv layers in models.

#430 sgsdxzy closed 1 week ago
0
[Installation]: Cannot install the library

#429 uysalfurkan closed 1 week ago
0
feat: LM Format Enforcer support

#428 AlpinDale closed 1 week ago
0
Port sampler+metadata changes from main to dev

#427 50h100a closed 1 week ago
0
[Usage]: odd use of GPUS number and tensor parallelism

#426 puppetm4st3r closed 1 week ago
2
[Feature]: Provide configuration via env vars or a configuration file

#425 alexandreteles opened 1 week ago
0
Fix: kobold api /tokencount

#424 Krovius closed 1 week ago
0
Split the exl2 weight ASAP.

#423 sgsdxzy closed 1 week ago
2
[Bug]: Mixtral-8x22b-instruct not running with AWQ

#421 SalomonKisters closed 1 week ago
10
Support sharded ggufs.

#420 sgsdxzy closed 1 week ago
0
[WIP] feat: Mamba support

#419 AlpinDale opened 2 weeks ago
1
[Feature]: Support hqq quantize method.

#418 Minami-su opened 2 weeks ago
0
[Bug]: gguf loading failed. config.json?

#417 juud79 opened 2 weeks ago
4
[Bug]: manually setting --max-model-len flag always leads to OOM, even if it is set very low

#414 SalomonKisters closed 2 weeks ago
2
[Feature]: Is there a reason CUDA 6.1 is the minimum? Would CUDA 6.0 on the P100 not work?

#413 Nero10578 opened 2 weeks ago
5
Fix max_num_batched_tokens for chunked_prefill.

#412 sgsdxzy closed 2 weeks ago
0
[Bug]: Converting gguf to state_dict

#411 heungson opened 2 weeks ago
3
Allow setting config-path when converting ggufs.

#410 sgsdxzy closed 2 weeks ago
0
Support twe lm_head for quantized weights.

#409 sgsdxzy closed 2 weeks ago
0
feat: EETQ

#408 AlpinDale closed 2 weeks ago
0
feat: Triton flash attention backend for ROCm

#407 AlpinDale closed 2 weeks ago
0
feat: add chunked prefill scheduler

#406 AlpinDale closed 2 weeks ago
0
feat: FP8 E4M3 KV Cache

#405 AlpinDale closed 3 weeks ago
0
Improve cohere model.

#404 sgsdxzy closed 3 weeks ago
0
feat: Intel CPU support

#403 AlpinDale closed 3 weeks ago
0
Speculative Decoding Part 4: Lookahead scheduling

#402 AlpinDale closed 3 weeks ago
0
[Crash]: Program gets terminated

#401 DuckY-Y opened 3 weeks ago
1
[Installation]: No module named 'aphrodite._C'

#400 DuckY-Y closed 3 weeks ago
2
[Bug]: served-model-name is unused

#399 mrseeker opened 3 weeks ago
1
feat: optimized layernorm kernels

#398 AlpinDale closed 3 weeks ago
0