Vinkal 72b24d96c6 model : make minicpm embedding_scale, residual_scale and logit_scale optional with legacy defaults (#16273) ... * minicpm: make GGUF scaling keys optional with legacy defaults Older MiniCPM GGUFs do not include the scaling metadata keys (minicpm.embedding_scale, minicpm.residual_scale, minicpm.logit_scale). The loader currently treats these as required, so quantization fails with: key not found in model: minicpm.embedding_scale This change restores backward compatibility by treating these keys as optional in the loader and using the older MiniCPM scaling values: embedding_scale = 12.0f residual_scale = 1.4f / sqrt(n_layer) logit_scale = 256.0f / n_embd When the GGUF provides the keys, their values override the defaults; otherwise the legacy defaults are used. Newer GGUFs that already include these keys are unaffected. Fixes: #16192 Signed-off-by: Vinkal Chudgar <vinkal.chudgar@gmail.com> * Update src/llama-model.cpp Committed as suggested. Thanks! Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> --------- Signed-off-by: Vinkal Chudgar <vinkal.chudgar@gmail.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>		2025-09-26 23:28:29 +02:00
..
CMakeLists.txt	kv-cache : drop the "unified" prefix (#15467)	2025-08-21 17:00:33 +03:00
llama-adapter.cpp	aLoRA Support (#15327)	2025-09-05 17:32:39 -06:00
llama-adapter.h	aLoRA Support (#15327)	2025-09-05 17:32:39 -06:00
llama-arch.cpp	model : add GroveMoE support (#15510)	2025-09-25 19:50:28 +02:00
llama-arch.h	model : add GroveMoE support (#15510)	2025-09-25 19:50:28 +02:00
llama-batch.cpp	perplexity : provide a helpful hint for has_cpl case in split_equal error. (#15304)	2025-08-14 14:03:30 +03:00
llama-batch.h
llama-chat.cpp	model : add grok-2 support (#15539)	2025-09-14 23:00:59 +02:00
llama-chat.h	model : add grok-2 support (#15539)	2025-09-14 23:00:59 +02:00
llama-context.cpp	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-context.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-cparams.cpp
llama-cparams.h	llama : bump max seq limit from 64 to 256 (#15916)	2025-09-18 12:47:56 +03:00
llama-grammar.cpp
llama-grammar.h
llama-graph.cpp	model : add GroveMoE support (#15510)	2025-09-25 19:50:28 +02:00
llama-graph.h	llama : add support for qwen3 reranker (#15824)	2025-09-25 11:53:09 +03:00
llama-hparams.cpp	kv-cache : fix SWA checks + disable cacheless iSWA (#15811)	2025-09-05 10:39:22 +03:00
llama-hparams.h	model : add GroveMoE support (#15510)	2025-09-25 19:50:28 +02:00
llama-impl.cpp
llama-impl.h	llama: use FA + max. GPU layers by default (#15434)	2025-08-30 16:32:10 +02:00
llama-io.cpp
llama-io.h
llama-kv-cache-iswa.cpp	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-kv-cache-iswa.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-kv-cache.cpp	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-kv-cache.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-kv-cells.h	llama : remove KV cache defragmentation logic (#15473)	2025-08-22 12:22:13 +03:00
llama-memory-hybrid.cpp	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-memory-hybrid.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-memory-recurrent.cpp	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-memory-recurrent.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-memory.cpp
llama-memory.h	llama: print memory breakdown on exit (#15860)	2025-09-24 16:53:48 +02:00
llama-mmap.cpp
llama-mmap.h
llama-model-loader.cpp	nvidia nemotron nano v2 (nemotronh) (#15507)	2025-08-28 18:39:31 -06:00
llama-model-loader.h	model: support GLM 4.5 family of models (#14939)	2025-08-04 20:29:25 +02:00
llama-model-saver.cpp
llama-model-saver.h
llama-model.cpp	model : make minicpm embedding_scale, residual_scale and logit_scale optional with legacy defaults (#16273)	2025-09-26 23:28:29 +02:00
llama-model.h	model : add GroveMoE support (#15510)	2025-09-25 19:50:28 +02:00
llama-quant.cpp	llama-quant : fix the verification of attention layers for encoder-decoder models (#16023)	2025-09-17 09:30:55 +02:00
llama-quant.h
llama-sampling.cpp	sampling : optimize dist sampler (#15704)	2025-09-03 18:16:26 +03:00
llama-sampling.h
llama-vocab.cpp	devops: add s390x & ppc64le CI (#15925)	2025-09-27 02:03:33 +08:00
llama-vocab.h	model : add grok-2 support (#15539)	2025-09-14 23:00:59 +02:00
llama.cpp	ggml-backend : add GGML_BACKEND_DEVICE_TYPE_IGPU device type (#15797)	2025-09-11 22:47:38 +02:00
unicode-data.cpp
unicode-data.h
unicode.cpp
unicode.h	devops: add s390x & ppc64le CI (#15925)	2025-09-27 02:03:33 +08:00