Search: “llm”

What's Changed app: fix ChatGPT model selector spacing mlxrunner: Evict prefix cache snapshots from the active conversation mlxrunner: check system free memory and wait for evicted runners before load

context : do not re-reserve the scheduler when toggling causal_attn (#28751) context : do not re-reserve the scheduler when toggling causal_attn llama_context::set_causal_attn() marks the scheduler to

common : add LLM-jp-4.1 Harmony dialect handler (#29681) LLM-jp-4.1 uses the GPT-OSS format, but its tokenizer decodes a space after every special token and parallel tool calls are separated by <|end|

Overview llama.cpp v0.6.0 introduces the new llama_batch_ext extended batch API (with llama_process) for mixed token/embedding inputs and MTP/deepstack state embeddings, adds support for the GLM-5.3-F

llama : fix unexpected graph reallocation in the k-pool models (#29958) llama : fix unexpected graph reallocation in the k-pool models Both k-pool models built a graph shape that depends on state the

llama: support both embd + raw tokens in batch (#29622) llama: support both embd + raw tokens in batch add to test-llama-archs also check case llm_arch_supports_mixed_batch = false constant graph topo

vocab : implement PLaMo-3 tokenizer pre-segmentation (#30045) vocab : implement PLaMo-3 tokenizer pre-segmentation The PLaMo-3 tokenizer inserts hard boundaries before running the Unigram DP, around <

models : consolidate nextn row cropping into shared helpers (#30017) mimo2 : always emit h_nextn the other nextn-capable models set it unconditionally Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-MOPD mo

The no-cache path builds a [n_tokens, n_tokens] mask and passes K and V unpadded, so FLASH_ATTN_EXT sees K->ne[1] == n_tokens. The CUDA kernels for head size 512 require a multiple of FATTN_KQ_STRIDE,

llama.cpp enables LLM inference in C/C++ , and release b9014 represents a build revision or commit snapshot from the llama.cpp repository on GitHub. This release would include bug fixes, feature impro

b9523 is a release build of llama.cpp, an open-source tool for LLM inference in C/C++ that enables language model execution with minimal setup on a wide range of hardware locally and in the cloud. The

llama.cpp build b8781 is the latest release of the open-source C/C++ LLM inference framework, published on April 13, 2026. Its primary change introduces a dedicated DeepSeek v3.2 chat parser along wit

llama.cpp enables LLM inference in C/C++ , and release b9014 represents a build revision or commit snapshot from the llama.cpp repository on GitHub. This release would include bug fixes, feature impro

b9414 is a release build of llama.cpp that includes improvements to CUDA PTX version checking , which helps prevent incorrect kernel dispatch on different GPU architectures. This build also adds suppo

b9523 is a release build of llama.cpp, an open-source tool for LLM inference in C/C++ that enables language model execution with minimal setup on a wide range of hardware locally and in the cloud. The

mtmd: Add Vision Support for Minimax-M3 (#25113) Add preliminary MiniMax-M3 support Text-only port that re-uses existing components: MiniMax-M2 style GQA with per-head QK-norm and partial rotary, Deep

model: add Qwen3.8-Flash-Next (qwen4exp) (#27742) gguf: add qwen4exp (Qwen3.8-Flash-Next) arch and converter Adds the GGUF-side plumbing for HF model_type qwen4_exp: MODEL_ARCH.QWEN4EXP plus tensors f

llama.cpp build **b8751** is a tagged release of the [ggml-org/llama.cpp](https://github.com/ggml-org/llama.cpp) project, a C/C++ library enabling high-performance LLM inference with minimal setup ...

Release b8833 of llama.cpp includes updates to the ggml-webgpu backend, fixing compiler warnings and refactoring FlashAttention encoding, along with workflow improvements and precision adjustments for

B9816 is a release build number for llama.cpp, an open-source project that enables efficient large language model inference in C/C++ on consumer hardware. This release likely contains bug fixes, perfo

mtmd: Add Vision Support for Minimax-M3 (#25113) Add preliminary MiniMax-M3 support Text-only port that re-uses existing components: MiniMax-M2 style GQA with per-head QK-norm and partial rotary, Deep

model-saver : fix expert shared/chunk FFN length key clobber (#26693) The saver called add_kv with LLM_KV_EXPERT_SHARED_FEED_FORWARD_LENGTH twice, the second time passing n_ff_chexp. gguf_set_val_u32

model : add support for HrmTextForCausalLM (DFM Mimir 1B) (#27625) model : add support for HrmTextForCausalLM (DFM Mimir 1B) HRM-Text runs two transformer stacks (low, high) in an alternating cycle ov

Overview llama.cpp v0.6.0 introduces the new llama_batch_ext extended batch API (with llama_process) for mixed token/embedding inputs and MTP/deepstack state embeddings, adds support for the GLM-5.3-F

The no-cache path builds a [n_tokens, n_tokens] mask and passes K and V unpadded, so FLASH_ATTN_EXT sees K->ne[1] == n_tokens. The CUDA kernels for head size 512 require a multiple of FATTN_KQ_STRIDE,