Fabio
c36510c21e
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
71 lines
2.1 KiB
C++
71 lines
2.1 KiB
C++
#pragma once
|
|
|
|
#include "engine/framework/assets/resource_bundle.h"
|
|
#include "engine/framework/assets/tensor_source.h"
|
|
#include "engine/framework/tokenizers/hf_tokenizer_json.h"
|
|
|
|
#include <cstdint>
|
|
#include <filesystem>
|
|
#include <memory>
|
|
#include <string>
|
|
#include <unordered_map>
|
|
#include <vector>
|
|
|
|
namespace engine::models::nemotron_asr {
|
|
|
|
struct NemotronFrontendConfig {
|
|
int64_t sample_rate = 16000;
|
|
int64_t feature_size = 128;
|
|
int64_t n_fft = 512;
|
|
int64_t win_length = 400;
|
|
int64_t hop_length = 160;
|
|
float preemphasis = 0.97f;
|
|
float log_zero_guard = 5.9604644775390625e-8f;
|
|
};
|
|
|
|
struct NemotronFastConformerEncoderConfig {
|
|
int64_t hidden_size = 1024;
|
|
int64_t intermediate_size = 4096;
|
|
int64_t layers = 24;
|
|
int64_t heads = 8;
|
|
int64_t kv_heads = 8;
|
|
int64_t conv_kernel = 9;
|
|
int64_t sliding_window = 57;
|
|
int64_t subsampling_factor = 8;
|
|
int64_t subsampling_channels = 256;
|
|
int64_t subsampling_kernel = 3;
|
|
int64_t subsampling_stride = 2;
|
|
int64_t max_position_embeddings = 5000;
|
|
int64_t default_lookahead_tokens = 3;
|
|
std::vector<int64_t> supported_lookahead_tokens;
|
|
};
|
|
|
|
struct NemotronConfig {
|
|
std::string model_type;
|
|
int64_t vocab_size = 13088;
|
|
int64_t blank_token_id = 13087;
|
|
int64_t pad_token_id = 0;
|
|
int64_t default_prompt_id = 101;
|
|
int64_t decoder_hidden_size = 640;
|
|
int64_t decoder_layers = 2;
|
|
int64_t max_symbols_per_step = 10;
|
|
int64_t num_prompts = 128;
|
|
int64_t prompt_intermediate_size = 2048;
|
|
NemotronFrontendConfig frontend;
|
|
NemotronFastConformerEncoderConfig encoder;
|
|
std::unordered_map<std::string, int64_t> prompt_dictionary;
|
|
};
|
|
|
|
struct NemotronASRAssets {
|
|
assets::ResourceBundle resources;
|
|
NemotronConfig config;
|
|
std::shared_ptr<const assets::TensorSource> source;
|
|
std::shared_ptr<engine::tokenizers::HuggingFaceTokenizerJson> tokenizer;
|
|
std::vector<uint8_t> special_token_ids;
|
|
std::string metaspace_replacement;
|
|
bool trim_leading_space = false;
|
|
};
|
|
|
|
std::shared_ptr<const NemotronASRAssets> load_nemotron_asr_assets(const std::filesystem::path & model_path);
|
|
|
|
} // namespace engine::models::nemotron_asr
|