Commits · Luigi/ZeroGPU-LLM-Inference

bugfixc: not using pipeline for response generation

939895d

Luigi commited on Apr 22

Add original SmolLM2 135M Instruct for comparaison

423dc1a

Luigi commited on Apr 22

Add SmolLM2-135M-Instruct-TaiwanChat

38fcc03

Luigi commited on Apr 22

Add SmolLM2-135M TaiwanChat

0d642b7

Luigi commited on Apr 22

default to gemma-3-4b

88a6a62

Luigi commited on Apr 12

model repo_id typo fix

89372fa

Luigi commited on Apr 12

enable web search by default

6235e63

Luigi commited on Apr 12

remove tinyllama which has bad response quality

a22cf42

Luigi commited on Apr 12

make streaming response

5ea073d

Luigi commited on Apr 12

apply history flatten before it goint to prompt

ef361b0

Luigi commited on Apr 12

better management on system prompt

5f6306a

Luigi commited on Apr 12

usue chat pipeline instead of model and tokenizer individually

ac8e9cc

Luigi commited on Apr 12

bugfix to padding-related issues

f248fec

Luigi commited on Apr 12

add attention mask

b6b3940

Luigi commited on Apr 12

Clean model description

4731160

Luigi commited on Apr 12

pin torch to 2.4.0

4c6b4c5

Luigi commited on Apr 12

enable zerogpu with decorator

c5b2897

Luigi commited on Apr 12

Apply ZeroGPU

d181b45

Luigi commited on Apr 12

switch to gradio version for stability reason

a703203

Luigi commited on Apr 12

add taiwan tinyllama

794ee70

Luigi commited on Apr 11

Improve responsiveness by asynchronously retrieving web search context

acda3f1

Luigi commited on Apr 11

add llama 3.2 taiwan 3b

b1544e2

Luigi commited on Apr 11

add minicpm3 4b

f5c0811

Luigi commited on Apr 11

increase xt length to max

629495e

Luigi commited on Apr 11

remove all moe

fafc8cb

Luigi commited on Apr 11

remove qwen 1.5 moe

6735035

Luigi commited on Apr 11

adjust title style

e9559bd

Luigi commited on Apr 11

use another version of qwen 1.5 moe

96e60d6

Luigi commited on Apr 11

add Qwen1.5-MoE

e17afaf

Luigi commited on Apr 11

Qwen2.5-MOE-6x1.5B

5eca666

Luigi commited on Apr 11

remove under 3b models

617be26

Luigi commited on Apr 11

Add model caching

d33dfcd

Luigi commited on Apr 11

UI/UX Improvement

eb215ff

Luigi commited on Apr 11

reset timeout timer once a new token is generated

35943b1

Luigi commited on Apr 11

open web search settgins to user

c9fd924

Luigi commited on Apr 10

add 2 more models

f7a541f

Luigi commited on Apr 10

apply new settings on duckduck search

d9421eb

Luigi commited on Apr 10

tune llama paramters

20484f3

Luigi commited on Apr 10

increase max_chars_per_result to 600

1155897

Luigi commited on Apr 10

increase max results to 6 for better web search

0c2fe1d

Luigi commited on Apr 10

increase ctx lenght to 2k

9ba47d1

Luigi commited on Apr 10

increase timeout to 5min

71d28c5

Luigi commited on Apr 10

Code simplification

248f5a7

Luigi commited on Apr 10

Enable speculattive decoding

a7fdfe6

Luigi commited on Apr 10

fix role disorder error in history

06a162a

Luigi commited on Apr 10

Add internet search feature

4e60755

Luigi commited on Apr 10

fix reasonning model's thought process display

9d3ca6c

Luigi commited on Apr 10

bugfix

4522453

Luigi commited on Apr 10

fix for multi-turn conv.

6c77ec7

Luigi commited on Apr 10

bugfix for think tag handling

14564aa

Luigi commited on Apr 10

Commit History

bugfixc: not using pipeline for response generation 939895d

Add original SmolLM2 135M Instruct for comparaison 423dc1a

Add SmolLM2-135M-Instruct-TaiwanChat 38fcc03

Add SmolLM2-135M TaiwanChat 0d642b7

default to gemma-3-4b 88a6a62

model repo_id typo fix 89372fa

enable web search by default 6235e63

remove tinyllama which has bad response quality a22cf42

make streaming response 5ea073d

apply history flatten before it goint to prompt ef361b0

better management on system prompt 5f6306a

usue chat pipeline instead of model and tokenizer individually ac8e9cc

bugfix to padding-related issues f248fec

add attention mask b6b3940

Clean model description 4731160

pin torch to 2.4.0 4c6b4c5

enable zerogpu with decorator c5b2897

Apply ZeroGPU d181b45

switch to gradio version for stability reason a703203

add taiwan tinyllama 794ee70

Improve responsiveness by asynchronously retrieving web search context acda3f1

add llama 3.2 taiwan 3b b1544e2

add minicpm3 4b f5c0811

increase xt length to max 629495e

remove all moe fafc8cb

remove qwen 1.5 moe 6735035

adjust title style e9559bd

use another version of qwen 1.5 moe 96e60d6

add Qwen1.5-MoE e17afaf

Qwen2.5-MOE-6x1.5B 5eca666

remove under 3b models 617be26

Add model caching d33dfcd

UI/UX Improvement eb215ff

reset timeout timer once a new token is generated 35943b1

open web search settgins to user c9fd924

add 2 more models f7a541f

apply new settings on duckduck search d9421eb

tune llama paramters 20484f3

increase max_chars_per_result to 600 1155897

increase max results to 6 for better web search 0c2fe1d

increase ctx lenght to 2k 9ba47d1

increase timeout to 5min 71d28c5

Code simplification 248f5a7

Enable speculattive decoding a7fdfe6

fix role disorder error in history 06a162a

Add internet search feature 4e60755

fix reasonning model's thought process display 9d3ca6c

bugfix 4522453

fix for multi-turn conv. 6c77ec7

bugfix for think tag handling 14564aa

bugfixc: not using pipeline for response generation

939895d

Add original SmolLM2 135M Instruct for comparaison

423dc1a

Add SmolLM2-135M-Instruct-TaiwanChat

38fcc03

Add SmolLM2-135M TaiwanChat

0d642b7

default to gemma-3-4b

88a6a62

model repo_id typo fix

89372fa

enable web search by default

6235e63

remove tinyllama which has bad response quality

a22cf42

make streaming response

5ea073d

apply history flatten before it goint to prompt

ef361b0

better management on system prompt

5f6306a

usue chat pipeline instead of model and tokenizer individually

ac8e9cc

bugfix to padding-related issues

f248fec

add attention mask

b6b3940

Clean model description

4731160

pin torch to 2.4.0

4c6b4c5

enable zerogpu with decorator

c5b2897

Apply ZeroGPU

d181b45

switch to gradio version for stability reason

a703203

add taiwan tinyllama

794ee70

Improve responsiveness by asynchronously retrieving web search context

acda3f1

add llama 3.2 taiwan 3b

b1544e2

add minicpm3 4b

f5c0811

increase xt length to max

629495e

remove all moe

fafc8cb

remove qwen 1.5 moe

6735035

adjust title style

e9559bd

use another version of qwen 1.5 moe

96e60d6

add Qwen1.5-MoE

e17afaf

Qwen2.5-MOE-6x1.5B

5eca666

remove under 3b models

617be26

Add model caching

d33dfcd

UI/UX Improvement

eb215ff

reset timeout timer once a new token is generated

35943b1

open web search settgins to user

c9fd924

add 2 more models

f7a541f

apply new settings on duckduck search

d9421eb

tune llama paramters

20484f3

increase max_chars_per_result to 600

1155897

increase max results to 6 for better web search

0c2fe1d

increase ctx lenght to 2k

9ba47d1

increase timeout to 5min

71d28c5

Code simplification

248f5a7

Enable speculattive decoding

a7fdfe6

fix role disorder error in history

06a162a

Add internet search feature

4e60755

fix reasonning model's thought process display

9d3ca6c

bugfix

4522453

fix for multi-turn conv.

6c77ec7

bugfix for think tag handling

14564aa