vllm
[Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default).
#26414

Merged

[Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default). #26414

noooop merged 25 commits into vllm-project:main from noooop:embed_fp16

FP16 Embedding Base64

dce9edb7

mergify added frontend

DarkLight1337 commented on 2025-10-08

noooop changed the title ~~[Model] FP16 Embedding Base64~~ [Model] Support FP16 Embedding Base64 (Still uses fp32 by default). 214 days ago

noooop changed the title ~~[Model] Support FP16 Embedding Base64 (Still uses fp32 by default).~~ [Frontend] Support FP16 Embedding Base64 (Still uses fp32 by default). 214 days ago

Merge branch 'main' into embed_fp16

d6451fcb

fix

7c7d124a

+ test_base64_embed_dtype

91964b7f

Merge branch 'main' into embed_fp16

cdc8174e

mergify added documentation

+ examples

20524d65

noooop force pushed to 20524d65 213 days ago

+ Comment

0df211e2

Merge branch 'main' into embed_fp16

71af8fba

noooop marked this pull request as ready for review 213 days ago

noooop requested a review from

robertgshaw2-redhat 213 days ago

noooop requested a review from

simon-mo 213 days ago

noooop requested a review from

aarnphm 213 days ago

noooop requested a review from

NickLucche 213 days ago

noooop requested a review from

chaunceyjiang 213 days ago

chatgpt-codex-connector commented on 2025-10-09

maxdebayser commented on 2025-10-09

maxdebayser approved these changes on 2025-10-09

Update vllm/entrypoints/openai/serving_embedding.py

02181488

Merge branch 'main' into embed_fp16

47520e4b

rename embedding_embed_dtype_client.py

9ae9cc7c

noooop commented on 2025-10-09

+ pooling

144b1ae6

+ pooling

edb982d0

fix

ef22f1e7

fix

096d6d19

DarkLight1337 commented on 2025-10-10

Update vllm/entrypoints/openai/serving_embedding.py

22d30681

Update vllm/entrypoints/openai/protocol.py

24b160b7

fix

3da26d3d

DarkLight1337 added ready

noooop changed the title ~~[Frontend] Support FP16 Embedding Base64 (Still uses fp32 by default).~~ [Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default). 211 days ago

mergify added needs-rebase

conflicts

b02b12d6

Merge branch 'main' into embed_fp16

40547215

mergify removed needs-rebase

add back

86cb7e47

fix

99e54cae

fix

c1d6927b

fix

8f898df1

DarkLight1337 approved these changes on 2025-10-13

Merge branch 'main' into embed_fp16

fa54a71d

noooop enabled auto-merge (squash) 209 days ago

noooop merged d2a79385 into main 209 days ago

noooop deleted the embed_fp16 branch 207 days ago

Reviewers

DarkLight1337

maxdebayser

Isotr0py

chatgpt-codex-connector

robertgshaw2-redhat

simon-mo

aarnphm

NickLucche

chaunceyjiang

Assignees

No one assigned

Labels

documentation frontend ready

Milestone

No milestone

vllm [Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default). #26414 Merged

[Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default). #26414

vllm
[Frontend][1/N] Improve all pooling task | Support FP16 Embedding Base64 (Still uses fp32 by default).
#26414

Merged