bug fix: handle saving/loading null layers in recurrent memory #14675
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
handle saving/loading null layers in recurrent memory
ggerganov
reviewed
Jul 14, 2025
compilade
approved these changes
Jul 14, 2025
There was a problem hiding this comment.
Thanks @l3utterfly! I've tested this with a Jamba model and llama-save-load-state, and it was indeed failing before, and is fixed by this change.
I'll add a test case to #14139 (once I also add variants for hybrid models) to help automatically detecting this kind of regression with hybrid architectures in the future.
CISC
reviewed
Jul 14, 2025
| @@ -1005,6 +1017,9 @@ bool llama_memory_recurrent::state_read_data(llama_io_read_i & io, uint32_t cell | |||
| } else { | |||
| // For each layer, read the values for each cell (transposed) | |||
| for (uint32_t il = 0; il < n_layer; ++il) { | |||
| // skip null layers | |||
| if(s_l[il] == nullptr) continue; | |||
There was a problem hiding this comment.
Suggested change
| if(s_l[il] == nullptr) continue; | |
| if (s_l[il] == nullptr) continue; |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently saving loading kv cache of recurrent memory crashes because layers can be null.
This mainly applies to the new LiquidAI/LFM2 models.
Tested with: https://huggingface.co/LiquidAI/LFM2-350M-GGUF