aha 0.2.5

aha model inference library, now supports Qwen(2.5VL/3/3VL/3.5/ASR/3Embedding/3Reranker), MiniCPM4, VoxCPM/1.5, DeepSeek-OCR/2, Hunyuan-OCR, PaddleOCR-VL/1.5, RMBG2.0, GLM(ASR-Nano-2512/OCR), Fun-ASR-Nano-2512, LFM(2/2.5/2VL/2.5VL)
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61

use candle_nn::Activation;

#[derive(Debug, Clone, PartialEq, serde::Deserialize)]
pub struct W2VBert2_0Config {
    pub activation_dropout: f32,
    pub adapter_act: String,
    pub adapter_kernel_size: usize,
    pub adapter_stride: usize,
    pub add_adapter: bool,
    pub apply_spec_augment: bool,
    pub attention_dropout: f32,
    pub bos_token_id: usize,
    pub classifier_proj_size: usize,
    pub codevector_dim: usize,
    pub conformer_conv_dropout: f32,
    pub contrastive_logits_temperature: f32,
    pub conv_depthwise_kernel_size: usize,
    pub ctc_loss_reduction: String,
    pub ctc_zero_infinity: bool,
    pub diversity_loss_weight: f32,
    pub eos_token_id: usize,
    pub feat_proj_dropout: f32,
    pub feat_quantizer_dropout: f32,
    pub feature_projection_input_dim: usize,
    pub final_dropout: f32,
    pub hidden_act: Activation,
    pub hidden_dropout: f32,
    pub hidden_size: usize,
    pub initializer_range: f32,
    pub intermediate_size: usize,
    pub layer_norm_eps: f64,
    pub layerdrop: f32,
    pub left_max_position_embeddings: usize,
    pub mask_feature_length: usize,
    pub mask_feature_min_masks: usize,
    pub mask_feature_prob: f32,
    pub mask_time_length: usize,
    pub mask_time_min_masks: usize,
    pub mask_time_prob: f32,
    pub max_source_positions: usize,
    pub num_adapter_layers: usize,
    pub num_attention_heads: usize,
    pub num_codevector_groups: usize,
    pub num_codevectors_per_group: usize,
    pub num_hidden_layers: usize,
    pub num_negatives: usize,
    pub output_hidden_size: usize,
    pub pad_token_id: usize,
    pub position_embeddings_type: String,
    pub proj_codevector_dim: usize,
    pub right_max_position_embeddings: usize,
    pub rotary_embedding_base: usize,
    pub tdnn_dilation: Vec<usize>,
    pub tdnn_dim: Vec<usize>,
    pub tdnn_kernel: Vec<usize>,
    pub torch_dtype: String,
    pub use_intermediate_ffn_before_adapter: bool,
    pub use_weighted_layer_sum: bool,
    pub vocab_size: Option<usize>,
    pub xvector_output_dim: usize,
}