34class QBRUNTIME_EXPORT ModelVariantHandle {
36 ModelVariantHandle(
const ModelVariantHandle& other) =
delete;
37 ModelVariantHandle(ModelVariantHandle&& other) =
delete;
38 ModelVariantHandle& operator=(
const ModelVariantHandle& rhs) =
delete;
39 ModelVariantHandle& operator=(ModelVariantHandle&& rhs)
noexcept =
delete;
40 ~ModelVariantHandle();
129 std::vector<Buffer> acquireInputBuffer(
130 const std::vector<std::vector<int>>& seqlens = {})
const;
131 std::vector<Buffer> acquireOutputBuffer(
132 const std::vector<std::vector<int>>& seqlens = {})
const;
133 std::vector<std::vector<Buffer>> acquireInputBuffers(
134 int batch_size,
const std::vector<std::vector<int>>& seqlens = {})
const;
135 std::vector<std::vector<Buffer>> acquireOutputBuffers(
136 int batch_size,
const std::vector<std::vector<int>>& seqlens = {})
const;
139 StatusCode releaseBuffer(std::vector<Buffer>& buffer)
const;
140 StatusCode releaseBuffers(std::vector<std::vector<Buffer>>& buffers)
const;
143 StatusCode repositionInputs(
const std::vector<float*>& input,
144 std::vector<Buffer>& input_buf,
145 const std::vector<std::vector<int>>& seqlens = {})
const;
146 StatusCode repositionOutputs(
const std::vector<Buffer>& output_buf,
147 std::vector<float*>& output,
148 const std::vector<std::vector<int>>& seqlens = {})
const;
149 StatusCode repositionOutputs(
const std::vector<Buffer>& output_buf,
150 std::vector<std::vector<float>>& output,
151 const std::vector<std::vector<int>>& seqlens = {})
const;
152 StatusCode repositionInputs(
const std::vector<uint8_t*>& input,
153 std::vector<Buffer>& input_buf,
154 const std::vector<std::vector<int>>& seqlens = {})
const;
157 StatusCode repositionInputs(
const std::vector<float*>& input,
158 std::vector<std::vector<Buffer>>& input_buf,
159 const std::vector<std::vector<int>>& seqlens = {})
const;
160 StatusCode repositionOutputs(
const std::vector<std::vector<Buffer>>& output_buf,
161 std::vector<float*>& output,
162 const std::vector<std::vector<int>>& seqlens = {})
const;
163 StatusCode repositionOutputs(
const std::vector<std::vector<Buffer>>& output_buf,
164 std::vector<std::vector<float>>& output,
165 const std::vector<std::vector<int>>& seqlens = {})
const;
166 StatusCode repositionInputs(
const std::vector<uint8_t*>& input,
167 std::vector<std::vector<Buffer>>& input_buf,
168 const std::vector<std::vector<int>>& seqlens = {})
const;
172 ModelVariantHandle(
int variant_idx,
const ModelImpl& model_impl);
175 const ModelImpl& mModelImpl;
177 friend class ModelImpl;