Model_name
stringclasses 2
values | Train_size
int64 100
100
| Test_size
int64 10
10
| arg
dict | lora
listlengths 3
8
| Parameters
int64 409M
912M
| Trainable_parameters
int64 1.77M
161M
| r
int64 8
8
| Memory Allocation
stringclasses 2
values | Training Time
stringclasses 2
values | Performance
dict |
|---|---|---|---|---|---|---|---|---|---|---|
facebook/bart-large-mnli
| 100
| 10
|
{
"auto_find_batch_size": true,
"gradient_accumulation_steps": 4,
"learning_rate": 0.00005,
"logging_steps": 1,
"lr_scheduler_type": "linear",
"num_train_epochs": 1,
"optim": "adamw_8bit",
"output_dir": "outputs",
"per_device_train_batch_size": null,
"report_to": "none",
"save_strategy": "no",
"save_total_limit": 0,
"seed": 3407,
"warmup_steps": 5,
"weight_decay": 0.01
}
|
[
"k_proj",
"q_proj",
"v_proj"
] | 409,124,878
| 1,769,472
| 8
|
1950.61
|
14.32
|
{
"accuracy": 0.1,
"f1_macro": 0.027777777777777776,
"f1_weighted": 0.05,
"precision": 0.018518518518518517,
"recall": 0.05555555555555555
}
|
unsloth/Qwen3-0.6B-unsloth-bnb-4bit
| 100
| 10
|
{
"auto_find_batch_size": null,
"gradient_accumulation_steps": 4,
"learning_rate": 0.00002,
"logging_steps": 10,
"lr_scheduler_type": "linear",
"num_train_epochs": 1,
"optim": "adamw_8bit",
"output_dir": "outputs",
"per_device_train_batch_size": 8,
"report_to": "none",
"save_strategy": null,
"save_total_limit": null,
"seed": 3407,
"warmup_steps": 5,
"weight_decay": 0.01
}
|
[
"down_proj",
"gate_proj",
"k_proj",
"lm_head",
"o_proj",
"q_proj",
"up_proj",
"v_proj"
] | 912,261,120
| 160,628,736
| 8
|
3303.32
|
59.48
|
{
"accuracy": 0,
"f1_macro": 0,
"f1_weighted": 0,
"precision": 0,
"recall": 0
}
|
README.md exists but content is empty.
- Downloads last month
- 4