ParthSareen
|
a5d638dfe7
extras
|
hai 1 mes |
ParthSareen
|
4aeb67ef4c
sample: do all sorting in topK
|
hai 1 mes |
ParthSareen
|
3ba91634c1
sample: simplify top_k=0 sorting
|
hai 1 mes |
ParthSareen
|
1b7433b71e
sample: use container/heap for top_k
|
hai 1 mes |
Bruce MacDonald
|
a70820daa0
models/gemma3: remove final logit softcap (#9692)
|
hai 1 mes |
Shane-XB-Qian
|
6b45b1d6b4
cli: adding support ctrl-n/p like general cli (#9136)
|
hai 1 mes |
frob
|
b3af953a55
cli: don't exit for invalid model during /load. (#9576)
|
hai 1 mes |
Michael
|
ad4e0bf3be
Adding Gemma 3 to readme (#9671)
|
hai 1 mes |
Michael Yang
|
aee28501b5
Merge pull request #9661 from ollama/gemma
|
hai 1 mes |
jmorganca
|
83f0ec8269
all: address linter errors
|
hai 1 mes |
jmorganca
|
c6b6938b3a
kvcache: fix tests by adding AvgPool2D stub
|
hai 1 mes |
jmorganca
|
fb4664fcec
model: add more spm tokenizer tests
|
hai 1 mes |
jmorganca
|
20e3593863
model: validate left and right pairs before merging them
|
hai 1 mes |
Michael Yang
|
63a394068c
use 2d pooling
|
hai 1 mes |
Daniel Hiltgen
|
ab39e08eb9
llm: auto detect models that require Ollama Engine (#1)
|
hai 1 mes |
jmorganca
|
11bfa62796
add trailing \n\n after <end_of_image> to match reference implementation
|
hai 1 mes |
jmorganca
|
f63e62e546
reduce kernel size, add TODO for loading from config
|
hai 1 mes |
jmorganca
|
65b0f329d1
Revert "Allow models to force a new batch"
|
hai 1 mes |
Jesse Gross
|
06007c0a18
Allow models to force a new batch
|
hai 1 mes |
Jesse Gross
|
a8e83a7654
Disable causal attention based on batch index
|
hai 1 mes |
Jesse Gross
|
475005504e
Restrict Gemma to a single image per request
|
hai 1 mes |
Jesse Gross
|
2c40c4d35e
Fix follow up images and images split across batches
|
hai 1 mes |
Michael Yang
|
e95278932b
use non-causal mask only for image positions
|
hai 1 mes |
Michael Yang
|
9d2a20a763
use non-causal mask for inputs with images
|
hai 1 mes |
Patrick Devine
|
2e54d72fc3
fix gemma3 1b conversion
|
hai 1 mes |
Michael Yang
|
6b32a2d549
compat with upstream gguf
|
hai 1 mes |
Michael Yang
|
c5cbe4fc2a
fallback to cpu
|
hai 1 mes |
Michael Yang
|
f888912870
fix vision encoder
|
hai 1 mes |
Michael Yang
|
9e4642e9b3
ollama debug tensor
|
hai 1 mes |
Michael Yang
|
6b0486c216
duplicate token_embd to output
|
hai 1 mes |