DX-596-together-fine-tuning: sync with mintlify-docs#932#25
Merged
Merged
Conversation
…cs#932 Adds Max LoRA rank caps documentation to align with the new Max LoRA Rank column added in togethercomputer/mintlify-docs#932 (docs/fine-tuning-models.mdx). - references/supported-models.md: new 'Max LoRA rank caps' table listing the model families capped at 16 (Moonshot Kimi K2, Z.ai GLM-5/5.1, DeepSeek R1/V3 non-distill families) vs the default cap of 64. - references/deployment.md: lora_r row notes the per-model cap and links to supported-models.md.
sadreamer
approved these changes
Jun 4, 2026
| | `warmup_ratio` | float | 0 | Warmup step ratio | | ||
| | `lora` | bool | true | Use LoRA method | | ||
| | `lora_r` | int | 64 | LoRA rank | | ||
| | `lora_r` | int | 64 | LoRA rank. Per-model max -- 64 for most models; 16 for Moonshot Kimi-K2 family, Z.ai GLM-5/5.1, and DeepSeek R1/V3 (non-distill) families. See [supported-models.md](supported-models.md). | |
There was a problem hiding this comment.
Probably it's better to not specify any specific values and just redirect to supported-models.md for up-to-date info
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Syncs
together-fine-tuningskill with togethercomputer/mintlify-docs#932 (merged).Source PR
Triggering doc files
docs/fine-tuning-models.mdxSkill changes
references/supported-models.md: added a new Max LoRA rank caps table after the LoRA model tables, capturing the per-modellora_rceilings introduced by the new Max LoRA Rank column. Models capped at 16: Moonshot Kimi K2 family, Z.ai GLM-5/5.1, DeepSeek R1/V3 (non-distill) families. All other LoRA-supported models keep the default cap of 64.references/deployment.md: updated thelora_rrow in the training-parameters table to note that the rank is capped per model and link tosupported-models.mdfor the cap table.No SKILL.md, scripts, agents, or other references touched.
Generated by the Sync Skills Cursor Automation. Please review before merging.