|
ember.cpp 0.1.0
C++ inference engine for Llama-family and paperkiln models (CPU, CUDA, WebAssembly)
|

Go to the source code of this file.
Namespaces | |
| namespace | ember_ir |
| node __pad0__ |
Definition at line 9 of file graph_ir_schema.h.
| node add |
Definition at line 9 of file graph_ir_schema.h.
Referenced by microtorch::llama3::split_words().
| node affine |
Definition at line 9 of file graph_ir_schema.h.
| node allOf |
Definition at line 9 of file graph_ir_schema.h.
| node append |
Definition at line 9 of file graph_ir_schema.h.
| node attention |
Definition at line 9 of file graph_ir_schema.h.
| node conv1d_causal |
Definition at line 9 of file graph_ir_schema.h.
| node depth_attend |
Definition at line 9 of file graph_ir_schema.h.
| node depth_route |
Definition at line 9 of file graph_ir_schema.h.
| node expr |
Definition at line 9 of file graph_ir_schema.h.
| node gelu |
Definition at line 9 of file graph_ir_schema.h.
| node in |
Definition at line 9 of file graph_ir_schema.h.
| node layernorm |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::build_ir_from_gguf().
| node linear |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::build_ir_from_gguf().
| node linear_attention |
Definition at line 9 of file graph_ir_schema.h.
| node list |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::rec::Recurrence::compile().
| node mla_attention |
Definition at line 9 of file graph_ir_schema.h.
| node moe_ffn |
Definition at line 9 of file graph_ir_schema.h.
| node mul |
Definition at line 9 of file graph_ir_schema.h.
| node mul_col |
Definition at line 9 of file graph_ir_schema.h.
| node out |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::cpu::affine(), microtorch::llama3::append_utf8(), attention_kernel(), attention_optimized_det_kernel(), attention_optimized_kernel(), ember_ir::Interpreter::bind_tensor(), bpe_build_prompt(), ember::build_chat_prompt(), ember::build_general_prompt(), ember_ir::build_ir_from_gguf(), microtorch::llama3::byte_encode(), ember_ir::cpu::conv1d_causal_row(), microtorch::llama3::cpts_from_utf8(), ember_kvq::decode_cuda(), decode_embed_kernel(), ember::nfc::decompose(), ember_ir::cpu::depth_attend(), dequant_legacy_to_f32(), dequant_span_to_f32(), microtorch::bpe::Tokenizer::encode(), ember::QwenNfcTokenizer::encode(), microtorch::llama3::Tokenizer::encode(), ember::SpmTokenizer::encode(), microtorch::bpe::Tokenizer::encode_chunk(), ember::SpmTokenizer::encode_fragment(), microtorch::llama3::Tokenizer::encode_word(), ember_ir::es_number_string(), ember_ir::expr::eval(), microtorch::chat::exchanges(), ember::ReplyBuilder::finish(), ember_ir::Interpreter::forward(), ember_ir::Interpreter::forward_impl(), gelu_cpu(), gelu_new_cpu(), tinyllama::TinyLlamaSession::generate(), ember::insert_state_blocks(), ember_ir::ir_affine_cuda(), ember_ir::ir_binary_cuda(), ember_ir::ir_container_errors(), ember_ir::ir_conv1d_causal_cuda(), ember_ir::ir_depth_attend_cuda(), ember_ir::ir_document_errors(), ember_ir::ir_embedding_cuda(), ember_ir::ir_embedding_state_cuda(), ember_ir::ir_expr_cuda(), ember_ir::ir_gelu_cuda(), ember_ir::ir_layernorm_cuda(), ember_ir::ir_linear_attention_cuda(), ember_ir::ir_mla_attention_cuda(), ember_ir::ir_mul_col_cuda(), ember_ir::ir_pos_embedding_cuda(), ember_ir::ir_relu_cuda(), ember_ir::ir_rms_row_cuda(), ember_ir::ir_scale_cuda(), ember_ir::ir_schema_errors(), ember_ir::ir_sigmoid_cuda(), ember_ir::ir_silu_cuda(), ember_ir::ir_turn_state_attention_cuda(), ember_ir::ir_window_attention_cuda(), ember_ir::linear_attention_row_cpu(), main(), microtorch::chat::messages_from_json(), ember_ir::cpu::mla_attention(), moe_ffn_cuda(), moe_ffn_f32_cpu(), moe_ffn_routed_f32_cpu(), ember_ir::cpu::mul_col(), ember::ReplyStream::next(), ember::nfc::nfc(), ember::nfc::nfc_utf8(), ember::parse_score_body(), microtorch::bpe::Tokenizer::pretokenize(), PYBIND11_MODULE(), ember_ir::Interpreter::recurrent_tensors(), ember_ir::cpu::relu(), ember_ir::cpu::rms_row(), rmsnorm_apply_kernel(), rmsnorm_optimized_det_kernel(), rmsnorm_optimized_kernel(), rmsnorm_vector_cpu(), ember_ir::Interpreter::run_nodes(), ember_ir::cpu::scale(), tinyllama::TinyLlamaSession::score_rows(), ember_ir::sha256_hex_bytes(), ember_ir::cpu::sigmoid(), silu_cpu(), silu_kernel(), softmax_normalize_kernel(), softmax_optimized_kernel(), softmax_vector_cpu(), ember_ir::rec::Recurrence::step(), ember_ir::cpu::sub(), swiglu_kernel(), ember::detail::tool_definitions(), uint8_vector_to_uint16_vector(), ember::utf8_valid(), and weighted_sum_probs_v().
| node pos_embedding |
Definition at line 9 of file graph_ir_schema.h.
| node properties |
Definition at line 9 of file graph_ir_schema.h.
| node recurrence |
Definition at line 9 of file graph_ir_schema.h.
| node relu |
Definition at line 9 of file graph_ir_schema.h.
| node required |
Definition at line 9 of file graph_ir_schema.h.
| node rms_row |
Definition at line 9 of file graph_ir_schema.h.
| node rmsnorm |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::build_ir_from_gguf().
| node rope |
Definition at line 9 of file graph_ir_schema.h.
| node scale |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::cpu::affine(), attention_cuda(), attention_cuda_optimized(), attention_cuda_selective_dequant(), attention_decode_at_cuda(), attention_devpos_cuda(), attention_kernel(), attention_optimized_det_kernel(), attention_optimized_kernel(), attention_selective_dequant_kernel(), ember_ir::build_ir_from_gguf(), calculate_attention_scores(), ember_kvq::decode_cuda(), ember_ir::ir_affine_cuda(), ember_ir::ir_mla_attention_cuda(), ember_ir::ir_turn_state_attention_cuda(), ember_ir::ir_window_attention_cuda(), ember_ir::cpu::mla_attention(), quantize_q6_k(), ember_ir::Interpreter::run_nodes(), tinyllama::sample_top_k_top_p_temperature(), simd_scaled_add(), ember_tiled::tiled_split_block(), vec_dot_q4_k_q8_k_cpu(), and vec_dot_q6_k_q8_k_cpu().
| node selective_scan |
Definition at line 9 of file graph_ir_schema.h.
| node sigmoid |
Definition at line 9 of file graph_ir_schema.h.
Referenced by moe_ffn_routed_f32_cpu(), and ember_ir::Interpreter::run_nodes().
| node silu |
Definition at line 9 of file graph_ir_schema.h.
| node split |
Definition at line 9 of file graph_ir_schema.h.
Referenced by generate_bpe_merges_from_vocab_scores().
| node ssm_scan |
Definition at line 9 of file graph_ir_schema.h.
| node sub |
Definition at line 9 of file graph_ir_schema.h.
Referenced by ember_ir::device_supported(), ember_ir::has_state_tokens(), ember_ir::Interpreter::init(), quantize_q6_k(), and vec_dot_q4_k_q8_k_cpu().
| node then |
Definition at line 9 of file graph_ir_schema.h.