chat_llamar             Chat with a local model through an ellmer::Chat
                        object
chat_llamar_stop        Stop the server spawned by chat_llamar()
embed_llamar            Embedding provider for ragnar / standalone use
llama_apply_control_vector
                        Apply a control vector to a context
llama_backend_devices   List available backend devices
llama_batch_free        Free a llama batch allocated with
                        'llama_batch_init()'
llama_batch_init        Initialise a llama batch
llama_chat_apply_template
                        Apply chat template to messages
llama_chat_build        Build a tool-aware chat prompt and its parsing
                        grammar
llama_chat_builtin_templates
                        List built-in chat templates
llama_chat_parse        Parse raw model output into content and tool
                        calls
llama_chat_template     Get model's built-in chat template
llama_context_flash_attn
                        Flash attention a context actually uses
llama_detokenize        Detokenize token IDs back to text
llama_embed_batch       Batch embeddings for multiple texts
llama_embeddings        Extract embeddings for a text
llama_encode            Encode tokens using the encoder
                        (encoder-decoder models only)
llama_flash_attn_type_name
                        Name of a flash-attention type
llama_free_context      Free an inference context
llama_free_model        Free a loaded model
llama_gen_begin         Begin a streaming (token-by-token) generation
llama_gen_begin_at      Begin streaming generation from an
                        already-prefilled context
llama_gen_end           Finish a streaming generation
llama_gen_next          Pull the next chunk of a streaming generation
llama_generate          Generate text from a prompt
llama_generate_batch    Generate completions for multiple prompts in
                        parallel
llama_get_embeddings    Get all output token embeddings as a matrix
llama_get_embeddings_ith
                        Get embeddings for the i-th token in the batch
llama_get_embeddings_seq
                        Get pooled embeddings for a sequence
llama_get_logits        Get logits from the last decode step
llama_get_logits_ith    Get logits for a specific token position
llama_get_model         Get the model associated with a context
llama_get_verbosity     Get current verbosity level
llama_hf_cache_clear    Clear the model cache
llama_hf_cache_dir      Get the cache directory for downloaded models
llama_hf_cache_info     Show information about the model cache
llama_hf_download       Download a GGUF model from Hugging Face
llama_hf_list           List GGUF files in a Hugging Face repository
llama_image_eval        Evaluate an image + prompt into a llama context
llama_image_load        Load an image file into an mtmd bitmap
llama_load_model        Load a GGUF model file
llama_load_model_from_splits
                        Load a model split across several GGUF files
llama_load_model_hf     Load a model directly from Hugging Face
llama_lora_alora_invocation_tokens
                        Invocation tokens of an activated LoRA (aLoRA)
llama_lora_apply        Apply a LoRA adapter to context
llama_lora_clear        Remove all LoRA adapters from context
llama_lora_load         Load a LoRA adapter
llama_lora_meta         Read the metadata of a LoRA adapter
llama_lora_remove       Remove a LoRA adapter from context
llama_max_devices       Get maximum number of devices
llama_max_parallel_sequences
                        Get the maximum number of parallel sequences
llama_max_tensor_buft_overrides
                        Get the maximum number of tensor buffer-type
                        overrides
llama_memory_breakdown_print
                        Print memory breakdown by device
llama_memory_can_shift
                        Check if the KV cache supports shifting
llama_memory_clear      Clear the KV cache
llama_memory_seq_add    Shift token positions in a sequence
llama_memory_seq_cp     Copy a sequence in the KV cache
llama_memory_seq_div    Integer-divide token positions in a sequence
llama_memory_seq_keep   Keep only one sequence in the KV cache
llama_memory_seq_pos_range
                        Get position range for a sequence
llama_memory_seq_rm     Remove tokens from a sequence in the KV cache
llama_model_cls_labels
                        Classifier output labels
llama_model_decoder_start_token
                        Token that starts decoding in encoder-decoder
                        models
llama_model_info        Get model metadata
llama_model_meta        Get all model metadata as a named character
                        vector
llama_model_meta_val    Get a single model metadata value by key
llama_model_sampling_meta
                        Sampling parameters recommended by the model
                        author
llama_mtmd_load         Load a multimodal projector (mmproj)
llama_mtmd_marker       Media marker string for multimodal prompts
llama_mtmd_set_verbosity
                        Set verbosity of the multimodal subsystem
llama_mtmd_support_audio
                        Does this multimodal context support audio?
llama_mtmd_support_vision
                        Does this multimodal context support vision
                        (images)?
llama_n_batch           Get logical batch size
llama_n_ctx             Get context window size
llama_n_ctx_seq         Get per-sequence context window size
llama_n_seq_max         Get maximum number of sequences
llama_n_threads         Get number of threads for single-token
                        generation
llama_n_threads_batch   Get number of threads for batch processing
llama_n_ubatch          Get physical micro-batch size
llama_new_context       Create an inference context
llama_numa_init         Initialize NUMA optimization
llama_perf              Get performance statistics
llama_perf_print        Print performance statistics to the console
llama_perf_reset        Reset performance counters
llama_perf_sampler      Sampler performance statistics
llama_pooling_type      Get pooling type
llama_sampler_accept    Feed a token to a sampler
llama_sampler_chain_add
                        Add a sampler to a chain
llama_sampler_chain_from_params
                        Build a sampler chain from a parameter list
llama_sampler_chain_get
                        Get a sampler out of a chain
llama_sampler_chain_n   Number of samplers in a chain
llama_sampler_chain_new
                        Create an empty sampler chain
llama_sampler_chain_remove
                        Remove a sampler from a chain
llama_sampler_clone     Copy a sampler
llama_sampler_free      Free a sampler or chain
llama_sampler_get_seed
                        Seed used by a sampler
llama_sampler_name      Name of a sampler
llama_sampler_new       Create a single sampler
llama_sampler_params    Describe a sampler chain
llama_sampler_reset     Reset a sampler's internal state
llama_serve_anthropic   Serve an Anthropic Messages API-compatible
                        endpoint for a local model
llama_serve_openai      Serve an OpenAI-compatible HTTP API for a local
                        model
llama_set_abort_callback
                        Set or clear the abort callback
llama_set_causal_attn   Set causal attention mode
llama_set_threads       Set the number of threads for a context
llama_set_verbosity     Set logging verbosity level
llama_set_warmup        Set warmup mode
llama_split_path        Build the path of one chunk of a split GGUF
llama_split_prefix      Recover the prefix from a split GGUF path
llama_state_data        Copy the context state to and from raw bytes
llama_state_get_size    Get the size of the serialized context state in
                        bytes
llama_state_load        Load context state from file
llama_state_save        Save context state to file
llama_state_seq         Save and restore the state of a single sequence
llama_state_seq_file    Save and load the state of a single sequence to
                        a file
llama_supports_gpu      Check whether GPU offloading is available
llama_supports_mlock    Check whether memory locking is supported
llama_supports_mmap     Check whether memory-mapped file I/O is
                        supported
llama_supports_rpc      Check whether RPC backend is available
llama_synchronize       Synchronize asynchronous computation
llama_system_info       Get system information string
llama_time_us           Get current time in microseconds
llama_token_to_piece    Convert a single token ID to its text piece
llama_tokenize          Tokenize text into token IDs
llama_vocab_add_special
                        Tokenizer BOS / EOS / SEP insertion defaults
llama_vocab_get_attr    Get the attribute flags of a token
llama_vocab_get_score   Get the score of a token
llama_vocab_get_text    Get the text representation of a token
llama_vocab_info        Get vocabulary special token IDs
llama_vocab_is_control
                        Check if a token is a control token
llama_vocab_is_eog      Check if a token is an end-of-generation token
llama_vocab_special_tokens
                        Additional special token IDs
llama_vocab_type        Get vocabulary type
