Skip to content

Enhance rslora comment clarity across notebooks and scripts #144

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Sneakr wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Enhance rslora comment clarity across notebooks and scripts #144

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
751b058
Reapply comment clarity fixes after Gemini suggestions
Sneakr Dec 18, 2025
352c084
Fixed LoftQ comment
Sneakr Dec 18, 2025
be9b955
Remove loftq_update_log.txt (accidental commit with sensitive path)
Sneakr Dec 18, 2025
6c77f2f
Tweak comment wording for clarity
Sneakr Dec 22, 2025
087dc8d
Tweak comment wording for clarity, include lower ranks
Sneakr Dec 22, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
4 changes: 2 additions & 2 deletions nb/CodeForces-cot-Finetune_for_Reasoning_on_CodeForces.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -253,8 +253,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Dec 22, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The new comment is an improvement, but the pull request description mentions (If True, set alpha ~ sqrt(r) for correct scaling), which is more descriptive and helpful than the hardcoded range ~16-32. It explains the relationship between alpha and r, providing better guidance. It's recommended to use this more informative comment across all modified files.

Suggested change
" use_rslora = False, # If True, set alpha to ~16–32\n",
" use_rslora = False, # If True, set alpha ~ sqrt(r) for correct scaling\n",

" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/CodeGemma_(7B)-Conversational.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,8 +153,8 @@
" bias = \"none\", # Supports any, but = \"none\" is optimized\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Deepseek_OCR_(3B)-Eval.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -541,8 +541,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Deepseek_OCR_(3B)-Evaluation.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -694,8 +694,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Deepseek_OCR_(3B).ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -541,8 +541,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/ERNIE_4_5_21B_A3B_PT-Conversational.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -618,8 +618,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
2 changes: 1 addition & 1 deletion nb/ERNIE_4_5_VL_28B_A3B_PT_Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -875,7 +875,7 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False,\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" target_modules = [\n",
" \"q_proj\", \"k_proj\", \"v_proj\", \"o_proj\",\n",
" \"gate_proj\", \"up_proj\", \"down_proj\",\n",
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma2_(2B)-Alpaca.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,8 +314,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma2_(9B)-Alpaca.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -343,8 +343,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma3N_(4B)-Audio.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -416,8 +416,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = [\n",
" \"q_proj\", \"k_proj\", \"v_proj\", \"o_proj\",\n",
" \"gate_proj\", \"up_proj\", \"down_proj\",\n",
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma3N_(4B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -474,8 +474,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma3_(270M).ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -364,8 +364,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma3_(4B)-Vision-GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -448,8 +448,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" use_gradient_checkpointing = \"unsloth\", # Reduces memory usage\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
Expand Down
4 changes: 2 additions & 2 deletions nb/Gemma3_(4B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -396,8 +396,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Granite4.0.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,8 +196,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Granite4.0_350M.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -366,8 +366,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/HuggingFace Course-Gemma3_(4B)-Vision-GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -450,8 +450,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" use_gradient_checkpointing = \"unsloth\", # Reduces memory usage\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
Expand Down
4 changes: 2 additions & 2 deletions nb/HuggingFace Course-Qwen2_5_7B_VL_GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -619,8 +619,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" use_gradient_checkpointing = \"unsloth\", # Reduces memory usage\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
Expand Down
4 changes: 2 additions & 2 deletions nb/HuggingFace Course-Qwen3_VL_(8B)-Vision-GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -558,8 +558,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" use_gradient_checkpointing = \"unsloth\", # Reduces memory usage\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
Expand Down
4 changes: 2 additions & 2 deletions nb/HuggingFace Course-gpt-oss-(20B)_A100-GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -765,8 +765,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-CodeForces-cot-Finetune_for_Reasoning_on_CodeForces.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -253,8 +253,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-CodeGemma_(7B)-Conversational.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,8 +153,8 @@
" bias = \"none\", # Supports any, but = \"none\" is optimized\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Deepseek_OCR_(3B)-Eval.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -541,8 +541,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Deepseek_OCR_(3B)-Evaluation.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -694,8 +694,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Deepseek_OCR_(3B).ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -541,8 +541,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-ERNIE_4_5_21B_A3B_PT-Conversational.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -618,8 +618,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
2 changes: 1 addition & 1 deletion nb/Kaggle-ERNIE_4_5_VL_28B_A3B_PT_Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -875,7 +875,7 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False,\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" target_modules = [\n",
" \"q_proj\", \"k_proj\", \"v_proj\", \"o_proj\",\n",
" \"gate_proj\", \"up_proj\", \"down_proj\",\n",
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma2_(2B)-Alpaca.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,8 +314,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma2_(9B)-Alpaca.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -343,8 +343,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma3N_(4B)-Audio.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -416,8 +416,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = [\n",
" \"q_proj\", \"k_proj\", \"v_proj\", \"o_proj\",\n",
" \"gate_proj\", \"up_proj\", \"down_proj\",\n",
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma3N_(4B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -474,8 +474,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma3_(270M).ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -364,8 +364,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma3_(4B)-Vision-GRPO.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -441,8 +441,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" use_gradient_checkpointing = \"unsloth\", # Reduces memory usage\n",
" # target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Gemma3_(4B)-Vision.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -396,8 +396,8 @@
" lora_dropout = 0,\n",
" bias = \"none\",\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
" target_modules = \"all-linear\", # Optional now! Can specify a list if needed\n",
")"
]
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Granite4.0.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,8 +196,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
4 changes: 2 additions & 2 deletions nb/Kaggle-Granite4.0_350M.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -366,8 +366,8 @@
" # [NEW] \"unsloth\" uses 30% less VRAM, fits 2x larger batch sizes!\n",
" use_gradient_checkpointing = \"unsloth\", # True or \"unsloth\" for very long context\n",
" random_state = 3407,\n",
" use_rslora = False, # We support rank stabilized LoRA\n",
" loftq_config = None, # And LoftQ\n",
" use_rslora = False, # If True, set alpha to ~16–32\n",
" loftq_config = None, # We support LoftQ\n",
")"
]
},
Expand Down
Loading