github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12065] Does Ollama support --ignore-eos? #8013

New Issue
Closed
opened 2026-04-12 20:14:04 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-12 20:14:04 -05:00
Owner
Copy Link

Originally created by @DoubleRedX on GitHub (Aug 25, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12065

It seems that "num_predict" parameter is equivalent to restricting the upper bound of token generation. If I want to specify the length of the output token as 128, I neesd the cooperation of '--ignore-eos'.I found the corresonding parameter in llama.cpp. Does Ollama support it?

llama.cpp code

if (params.sampling.ignore_eos) {
    for (llama_token i = 0; i < llama_vocab_n_tokens(vocab); i++) {
        if (llama_vocab_is_eog(vocab, i)) {
            LOG_INF("%s: added %s logit bias = %f\n", __func__, common_token_to_piece(lctx, i).c_str(), -INFINITY);
            params.sampling.logit_bias.push_back({i, -INFINITY});
        }
    }
}
Originally created by @DoubleRedX on GitHub (Aug 25, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12065 It seems that "num_predict" parameter is equivalent to restricting the upper bound of token generation. If I want to specify the length of the output token as 128, I neesd the cooperation of '--ignore-eos'.I found the corresonding parameter in llama.cpp. Does Ollama support it? llama.cpp code if (params.sampling.ignore_eos) { for (llama_token i = 0; i < llama_vocab_n_tokens(vocab); i++) { if (llama_vocab_is_eog(vocab, i)) { LOG_INF("%s: added %s logit bias = %f\n", __func__, common_token_to_piece(lctx, i).c_str(), -INFINITY); params.sampling.logit_bias.push_back({i, -INFINITY}); } } }
GiteaMirror commented 2026-04-12 20:14:06 -05:00
Author
Owner
Copy Link

@DoubleRedX commented on GitHub (Aug 25, 2025):

runner/llamarunner/runner.go

	// if it's an end of sequence token, break
	if s.model.TokenIsEog(token) {
		// TODO (jmorganca): we should send this back
		// as it's important for the /api/generate context
		// seq.responses <- piece

		s.removeSequence(i, llm.DoneReasonStop)
		continue
	}

Found it. The cpp code doesn't take effect which shouble be ignore.

<!-- gh-comment-id:3220060997 --> @DoubleRedX commented on GitHub (Aug 25, 2025): runner/llamarunner/runner.go // if it's an end of sequence token, break if s.model.TokenIsEog(token) { // TODO (jmorganca): we should send this back // as it's important for the /api/generate context // seq.responses <- piece s.removeSequence(i, llm.DoneReasonStop) continue } Found it. The cpp code doesn't take effect which shouble be ignore.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#8013
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?