Skip to content
This repository has been archived by the owner on Oct 11, 2024. It is now read-only.

Upstream sync 2024 03 24 #143

Merged
merged 197 commits into from
Mar 26, 2024
Merged

Upstream sync 2024 03 24 #143

merged 197 commits into from
Mar 26, 2024

Conversation

robertgshaw2-redhat
Copy link
Collaborator

@robertgshaw2-redhat robertgshaw2-redhat commented Mar 24, 2024 •
edited
Loading

SUMMARY:

upstream merge (sync) up to af9e534
cleaned up various ruff issues
made many comments about where conflict typically emerge and how to fix them
fixed correctness tests
updated nvcc threads

TEST PLAN:
nightly

ronensc and others added 30 commits February 21, 2024 18:18
@ronensc
Update comment (vllm-project#2934)
d7f3964
@Maxusmusti
Added early stopping to completion APIs (vllm-project#2939)
5574081
@esmeetu
Migrate MistralForCausalLM to LlamaForCausalLM (vllm-project#2868)
344020c
@WoosukKwon
Use Llama RMSNorm custom op for Gemma (vllm-project#2974)
95529e3
@mspronesti
chore(vllm): codespell for spell checking (vllm-project#2820)
93dc5a2
@WoosukKwon
Optimize GeGLU layer in Gemma (vllm-project#2975)
fd5dcc5
@44670
[FIX] Fix a bug in initializing Yarn RoPE (vllm-project#2983)
c530e2c
@WoosukKwon
Remove Flash Attention in test env (vllm-project#2982)
6f32cdd
@ronensc
Include tokens from prompt phase in counter_generation_tokens (vllm…
4caf704 
…-project#2802)
@zhaoyang-star
Fix nvcc not found in vlm-openai image (vllm-project#2781)
57f0449
@WoosukKwon
[Fix] Fissertion on YaRN model len (vllm-project#2984)
f7c1234
@hmellor
Port metrics from aioprometheus to prometheus_client (vllm-projec…
ef978fe 
…t#2730)
@jlcmoore
Add LogProbs for Chat Completions in OpenAI (vllm-project#2918)
70f3e8e
@pcmoritz @cadedaniel
Optimize Triton MoE Kernel (vllm-project#2979)
cfc15a1 
Co-authored-by: Cade Daniel <[email protected]>
@WoosukKwon
[Minor] Remove gather_cached_kv kernel (vllm-project#3043)
d6e4a13
@esmeetu
[Minor] Remove unused config files (vllm-project#3039)
d9f726c
@esmeetu
Don't use cupy when enforce_eager=True (vllm-project#3037)
c1c0d00
@esmeetu
Fix stablelm (vllm-project#3038)
4dd6416
@dachengai @WoosukKwon
Support Orion model (vllm-project#2539)
48a8f4a 
Co-authored-by: zhangdacheng <[email protected]>
Co-authored-by: Woosuk Kwon <[email protected]>
@Jingru
fix get_ip error in pure ipv6 environment (vllm-project#2931)
2410e32
@WoosukKwon
[Minor] Fix type annotation in fused moe (vllm-project#3045)
4bd18ec
@dylanwhawk
Support logit bias for OpenAI API (vllm-project#3027)
e0ade06
@WoosukKwon
[Minor] Fix StableLMEpochForCausalLM -> StableLmForCausalLM (vllm-pro…
8b430d7 
…ject#3046)
@sighingnow
Enable GQA support in the prefix prefill kernels (vllm-project#3007)
71bcaf9 
Signed-off-by: Tao He <[email protected]>
@ElefHead
multi-lora documentation fix (vllm-project#3064)
a868310
@AllenDou
Restrict prometheus_client >= 0.18.0 to prevent errors when importing…
e46fa5d 
… pkgs (vllm-project#3070)
@liangfu
[Neuron] Support inference with transformers-neuronx (vllm-project#2569)
3b7178c
@WoosukKwon
Add LoRA support for Gemma (vllm-project#3050)
929b4f2
@chu-tianxiang
Add Support for 2/3/8-bit GPTQ Quantization Models (vllm-project#2330)
01a5d18
@jaywonchung
Fix: AttributeError in OpenAI-compatible server (vllm-project#3018)
a6d471c
mgoin
mgoin approved these changes Mar 25, 2024
View reviewed changes
Copy link
Member

@mgoin mgoin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nice job with all the sync comments, lgtm

@varun-sundar-rabindranath varun-sundar-rabindranath merged commit b335bf7 into main Mar 26, 2024
2 checks passed
@varun-sundar-rabindranath varun-sundar-rabindranath deleted the upstream-sync-2024-03-24 branch March 26, 2024 19:47
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Reviewers

@mgoin mgoin mgoin approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.