For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Python class

LinearLoRA

`LinearLoRA`

class max.nn.LinearLoRA(in_dim, out_dim, max_num_loras, max_lora_rank, dtype, device, has_lora_bias=False, name=None, quantization_encoding=None)

source

Bases: Module, SupportsLoRA

Applies a linear transformation and LoRA to input:

$y_l = (xA^T) @ B^T$ . $y = (xW^T + b) + y_l$

linear_layer = LinearLoRA(
    in_dim=256,
    out_dim=128,
    max_lora_rank=16,
    max_num_loras=100,
    dtype=dtype.float32,
    device=DeviceRef.GPU(),
    has_bias=True,
    has_lora_bias=True,
    name="lora_linear"
)

lora_ids: TensorValue # shape: [max_num_loras,]
lora_ranks: TensorValue # shape: [max_num_loras,]
input_row_offsets: TensorValue
linear_layer.set_lora_batch_info(lora_ids, lora_ranks, input_row_offsets)

input_tensor: TensorValue
output = linear_layer(input_tensor)

Parameters:

in_dim (int)
out_dim (int)
max_num_loras (int)
max_lora_rank (int)
dtype (DType)
device (DeviceRef)
has_lora_bias (bool)
name (str | None)
quantization_encoding (QuantizationEncoding | None)

`set_lora_batch_info()`

set_lora_batch_info(lora_ids, lora_ranks, lora_grouped_offsets, num_active_loras, lora_end_idx, batch_seq_len, lora_ids_kv, lora_grouped_offsets_kv)

source

Parameters:

lora_ids (TensorValue)
lora_ranks (TensorValue)
lora_grouped_offsets (TensorValue)
num_active_loras (TensorValue)
lora_end_idx (TensorValue)
batch_seq_len (TensorValue)
lora_ids_kv (TensorValue)
lora_grouped_offsets_kv (TensorValue)

Return type:

None

LinearLoRA​

set_lora_batch_info()​

`LinearLoRA`

`set_lora_batch_info()`