ollama

Commit Graph

Author	SHA1	Message	Date
Jesse Gross	7073600797	ggml: Reduce log level of "key not found" Most of the time this is not an error.	11 months ago
Daniel Hiltgen	b1c40138da	win: lint fix (#10571 )	11 months ago
Ashok Gelal	17466217e5	Hide empty terminal window (#8668 ) This hides the LlamaServer blank window when chatting outside of the terminal (say like with an app like Msty). This has no other side effects when invoking it the regular way.	11 months ago
Jeffrey Morgan	1703d1472e	server: fix panic when runner.Options is nil (#10566 )	11 months ago
Jeffrey Morgan	913905028b	all: fix cgo compiler warnings on windows (#10563 )	11 months ago
湛露先生	7e5c8eee5c	file close check and close. (#10554 ) Signed-off-by: zhanluxianshen <zhanluxianshen@163.com>	11 months ago
Daniel Hiltgen	6a74bba7e7	win: ensure ollama paths come first (#10549 ) For all search path env vars make sure our dirs are first to avoid potentially finding other incompatible libraries on the users system. Also fixes a minor build script glitch for windows rocm	11 months ago
Daniel Hiltgen	76ea735aaf	sched: logging improvements (#10550 ) This enhances our logging in the scheduler. The initial "waiting for server" log no longer claims an initial error state (now "not responding" which better reflects the actual state). Runners now have slog wiring to report more details about the runner, including PID.	11 months ago
aritra saha	dd1d4e99e7	readme: add llama 4 models (#10530 )	11 months ago
Jesse Gross	a6ef73f4f2	ggml: Fix race that resulted in "context canceled" when loading Successfully completing processing with an errgroup cancels the associated context. However, we also have a goroutine that is checking for cancelation of the context. As a result, there is a race where the goroutine can pick up the cancelation and report an error, replacing the sucessful error message. To avoid that, this replaces the goroutine with a cancelation check when we are reading files. This also has the advantage of stopping all reads relatively quickly on error and also ensuring that there are no outstanding I/O operations when we return in this case. The downside is that if a file read blocks forever (for example, over the network) then cancelation of the context effectively won't be honored. However, this is also true for other smaller files we read and the tensors are read in small chunks (128K), so it's consistent and better on balance overall.	11 months ago
Jesse Gross	c2f5d6662b	ollamarunner: Re-enable worst case graph preallocation. Worst case graph preallocation was disabled by `a27462b` "ollamarunner: Temporarily disable worst case graph preallocation" since it caused crashes with large batches when not using the GPU. This backports upstream llama.cpp commit f057808 "ggml: Don't assert fail when tensor data changes (#13222)", which fixes the underlying bug and allows reverting the previous workaround.	11 months ago
Harsh Nevse	57fb759f3c	readme: update link to langchain in community integrations (#10465 )	11 months ago
Jeffrey Morgan	8dd12c873d	llama: update to commit e1e8e099 (#10513 )	11 months ago
frob	e6d2d04121	image: add vision capability for projector-based models (#10509 ) Co-authored-by: Richard Lyons <frob@cloudstaff.com>	11 months ago
Jesse Gross	074bac8447	kvcache: Log batch size if we can't find a slot In some cases, we can't find a cache slot when using sliding window attention. It would be helpful in this (and other cases) to know what the batch size is. Bug #10127	11 months ago
Jesse Gross	8e8f2c6d67	ollamarunner: Fix memory leak when processing images The context (and therefore associated input tensors) was not being properly closed when images were being processed. We were trying to close them but in reality we were closing over an empty list, preventing anything from actually being freed. Fixes #10434	11 months ago
AliAhmedNada	938e8447e8	readme: add Jirapt project to community integrations (#10522 )	11 months ago
aritra saha	d5d5f0c445	readme: change granite3.2 to granite3.3 (#10525 ) Update the list for readme	11 months ago
Michael Yang	a7835c6716	fix: write gguf padding (#10510 ) * add gguf_test * fix padding padding was being added to offset but not to the running count	11 months ago
Devon Rifkin	ad3c7c9bda	strip out thinking tags in message history for qwen3 & r1 (#10490 ) * strip out thinking tags in message history for qwen3 & r1 This is in advance of "proper" support where we'll make reasoning configurable and we'll parse out thinking/reasoning tags and provide them to the caller. These models expect there to be no thinking tags in the message history, so this should improve quality * parse model names instead of hacky prefix check	11 months ago
Daniel Hiltgen	415c8fcc3d	Fix "Stopping..." scheduler hang (#10487 ) * Adjust initial scheduler refCount Ensure we only set the refCount on success * sched: fix lock order inversion deadlock Under certain race conditions, there was a scenario where the scheduler would get into a deadlock while trying to update free space information while a model was trying to unload.	11 months ago
Daniel Hiltgen	718eda1b3e	Narrow set of paths we load GGML from (#10485 ) Users may have other incompatible GGML installs on their systems. This will prevent us from trying to load them from the path.	11 months ago
Shahin R	421b7edeb4	readme: add link to lumina, a lightweight React frontend client (#10378 )	11 months ago
batuhankadioglu	7b68e254c2	all: update several golang.org/x packages (#10436 )	11 months ago
Daniel Hiltgen	7bec2724a5	integration: fix embedding tests error handling (#10478 ) The cleanup routine from InitServerconnection should run in the defer of the test case to properly detect failures and report the server logs	11 months ago
Jesse Gross	a27462b708	ollamarunner: Temporarily disable worst case graph preallocation When we later have a large batch running purely on a CPU, this results the error: GGML_ASSERT(talloc->buffer_id >= 0) Disabling this means that we will incrementally reallocate memory as the graph grows. Fixes #10410	11 months ago
crStiv	6bf0b8193a	readme: fix typos (#10399 )	11 months ago
Devon Rifkin	db428adbb8	Merge pull request #10468 from ollama/drifkin/num-parallel-1	11 months ago
Devon Rifkin	fe5b9bb21b	lower default num parallel to 2 this is in part to "pay" for #10452, which doubled the default context length. The combination isn't fully neutral though, because even though the old 4x2k limit and the new 2x4k limit are memory equivalent, the 1x fallback is larger with 4k	11 months ago
Devon Rifkin	6ec71d8fb6	Merge pull request #10452 from ollama/drifkin/4096-context-length config: update default context length to 4096	11 months ago
Devon Rifkin	44b466eeb2	config: update default context length to 4096	11 months ago
Devon Rifkin	a25f3f8260	Merge pull request #10451 from ollama/revert-10364-drifkin/context-length Revert "increase default context length to 4096"	11 months ago
Devon Rifkin	dd93e1af85	Revert "increase default context length to 4096 (#10364 )" This reverts commit `424f648632`.	11 months ago
Devon Rifkin	d2ee599dcf	load arrays with up to 1024 elements when estimating This mirrors the old behavior before #10382	11 months ago
Devon Rifkin	6ed8898590	ggml: fix crash for array head counts If it's an array, it uses the max value in the array If array values for head counts becomes more popular, we can consider a more invasive change like #10225 to calculate more accurate estimates. Fixes: #9984	11 months ago
Michael Yang	5cfc1c39f3	model: fix build (#10416 )	11 months ago
Michael Yang	f0ad49ea17	memory	11 months ago
Michael Yang	7ba9fa9c7d	fixes for maverick	11 months ago
Michael Yang	8bf11b84c1	chunked attention	12 months ago
Michael Yang	470af8ab89	connect vision to text	12 months ago
Michael Yang	178761aef3	image processing Co-authored-by: Patrick Devine <patrick@infrahq.com>	12 months ago
Michael Yang	f0c66e6dea	llama4	1 year ago
Michael Yang	54055a6dae	fix test	11 months ago
Michael Yang	340448d2d1	explicitly decode maxarraysize 1024	11 months ago
Michael Yang	ced7d0e53d	fix parameter count	11 months ago
Michael Yang	a0dba0f8ae	default slice values	11 months ago
Michael Yang	5e20b170a7	update comment	11 months ago
Michael Yang	d26c18e25c	fix token type	11 months ago
Michael Yang	8d376acc9b	zero means zero use a default of 1024 when asking for zero is confusing since most calls seem to assume 0 means do not ready any data	11 months ago
Michael Yang	dc1e81f027	convert: use -1 for read all	11 months ago

... 2 3 4 5 6 ...

4385 Commits (f371260e75b86d77ac22671c24881f696c88973c) All Branches Search

4385 Commits (f371260e75b86d77ac22671c24881f696c88973c)

All Branches