model : Support Limite 1B - Violetto (#29433)

* gguf-py : add Limite arch

Assisted-by: Claude Opus 5.5
Signed-off-by: Giuseppe Scrivano <giuseppe@scrivano.org>

* convert : add Limite support

Assisted-by: Claude Opus 5.5
Signed-off-by: Giuseppe Scrivano <giuseppe@scrivano.org>

* llama : add Limite support

Assisted-by: Claude Opus 5.5
Signed-off-by: Giuseppe Scrivano <giuseppe@scrivano.org>

---------

Signed-off-by: Giuseppe Scrivano <giuseppe@scrivano.org>
This commit is contained in:
Giuseppe Scrivano authored and GitHub committed 2026-10-29 10:13:41 +02:00
1 parent 41be0bdc98
commit abb76ab380
15 files changed
+598 -2

No files matched your search

+7
View File
@@ -333,6 +333,13 @@ static bool tensor_allows_quantization(const llama_model_quantize_params * param
// these are not too big so keep them as it is
quantize &= name.find("per_layer_model_proj") == std::string::npos;
// limite: very small gates and dense connection mixers
if (arch == LLM_ARCH_LIMITE) {
quantize &= name.find("attn_gate.weight") == std::string::npos;
quantize &= name.find("attn_ve_gate.weight") == std::string::npos;
quantize &= name.find("mudd_") == std::string::npos;
}
// do not quantize positional embeddings and token types (BERT)
quantize &= name != LLM_TN(arch)(LLM_TENSOR_POS_EMBD, "weight");
quantize &= name != LLM_TN(arch)(LLM_TENSOR_TOKEN_TYPES, "weight");