nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 36 Packages Projects Releases Wiki Activity

Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

Browse Source 
* kimi linear model implementation

* kimi linear convert_hf_to_gguf

* kimi linear constants.py tensor_mapping.py

* Kimi Linear ggml.h

* kimi linear ggml-cpu

* Kimi Linear ggml-cuda

* Kimi Linear ggml.c

* kimi linear src/llama

* remove "const int64_t n_seq_tokens = q->ne[2];" to get rid of unused variable warning

* remove type mismatch warning

* read MoE params

* removed some hard coded code

* removed all hard code

* use DeepseekV2 tokenizer

* removed unnecessary internal methods called by the old set_vocab of KimiLinear

* rewrite get_vocab for KimiLinear. Removed all kda_scan code

* removed all traces of kda_scan

* reduce OP count by 1 due to removal of kda_scan

* Move KIMI_LINEAR to llm_arch_is_hybrid to enable KV cache

* set n_embd_head_k/v to ensure kv cache works

* don't quantize conv1d of Kimi Linear

* Kimi Linear backend agnostic

* removed LOG_INFO

* naive chunking form implemented

* fixed some comments

* add Kimi-K2 specific tokens to be recognized as EOG

* build_kda_autoregressive is implemented to replace build_kda_recurrent for faster inference. sync'd to b7682

* replaced Akk and Aqk with mul_mat and clamp

* no clamp version

* Moved Aqk computation out of the loop

* fixed typo and split wkv_b into wk_b and wv_b

* MLA KV cache support

* fix trailing spaces

* moved const llama_model & model; around to follow qwen3next format and see if it cna pass the -Wunused-private-field error

* fix trailing whitespace

* removed traling whitespaces in empty line + make sure indentation is multiple of 4

* try to make lint happy

* remove blank lines to make lint happy

* removed at least blank line containing white space

* fixed flake8 complaints locally

* return ggml_tensor * pair in kda_autoregressive and kda_chunking as in ngxson's Qwen3Next improvement

* removed Kimi-Linear specific change that causes failure at server-windows

* removed private: from kimi_linear to make build checks happy

* removed unnecessary ggml_cont before ggml_reshape

* created static function causal_conv1d to abtract similar code for q/k/v

* merged dt_bias to SSM_DT. Do -exp(log_A) in convert_hf_to_gguf.py.

* reverted to original

* fixed find_hparam calls. Fixed e_score_correction_bias to use bias instead of weight. Removed all ssm_conv bias terms.

* remove DT_B from constants.py. remove one comment line in llama-model.cpp

* new class llm_graph_input_mem_hybrid_k to get around the new MLA change. switch the concat order of ggml_concat calls in kimi-linear.cpp to accommodate MLA changes. Removed support for exp_probs_b.weight

* remove ssm_o_norm_b

* remove ssm_o_norm_b

* changed hparams.kda_head_dim to hparams.n_embd_head_kda. added TODO comment for class llama_graph_mem_hybrid_k

* removed all ggml_cont b4 ggml_reshape_4d

* Whitespace

* replaced all hparams.get with find_hparams

* added new names for n_experts, n_experts_used and score_func in TextModel and removed their code in KimiLinear in convert_hf_to_gguf.py. Removed unnecessary ggml_cont and GGML_ASSERT in kimi-linear.cpp

* use is_mla to switch between different mem_hybrid types

* fixed logical errors in convert_hf_to_gguf.py pointed out by CISC

* removed if else for required parameters kv_lora_rank and qk_rope_head_dim

* add back ggml_cont for Vcur

* minor changes

* removed extra line in llama-vocab.cpp. Added back the comment in llama-graph.cpp

* f16 gguf cannot run without context length

* made a mistake of adding back n_ctx parsing

---------

Co-authored-by: Piotr Wilkin (ilintar) <piotr.wilkin@syndatis.com>
This commit is contained in:
ymcki
2026-02-06 11:39:58 +01:00
committed by GitHub
co-authored by Piotr Wilkin
parent 1946e46f4c
commit 3688c4f504
18 changed files with 1521 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-vocab.cpp
+28 -17
View File
@@ -1752,26 +1752,33 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
// read bpe merges and populate bpe ranks
const int merges_keyidx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_MERGES).c_str());
// Kimi-K2 uses custom tokenization without traditional BPE merges
const bool is_kimi_k2 = (tokenizer_pre == "kimi-k2");
if (merges_keyidx == -1) {
throw std::runtime_error("cannot find tokenizer merges in model file\n");
}
const int n_merges = gguf_get_arr_n(ctx, merges_keyidx);
for (int i = 0; i < n_merges; i++) {
const std::string word = gguf_get_arr_str(ctx, merges_keyidx, i);
//GGML_ASSERT(unicode_cpts_from_utf8(word).size() > 0);
std::string first;
std::string second;
const size_t pos = word.find(' ', 1);
if (pos != std::string::npos) {
first = word.substr(0, pos);
second = word.substr(pos + 1);
if (!is_kimi_k2) {
throw std::runtime_error("cannot find tokenizer merges in model file\n");
}
// Kimi-K2 doesn't need merges, skip
LLAMA_LOG_INFO("%s: Kimi-K2 tokenizer detected, skipping BPE merges\n", __func__);
} else {
const int n_merges = gguf_get_arr_n(ctx, merges_keyidx);
for (int i = 0; i < n_merges; i++) {
const std::string word = gguf_get_arr_str(ctx, merges_keyidx, i);
//GGML_ASSERT(unicode_cpts_from_utf8(word).size() > 0);
bpe_ranks.emplace(std::make_pair(first, second), i);
std::string first;
std::string second;
const size_t pos = word.find(' ', 1);
if (pos != std::string::npos) {
first = word.substr(0, pos);
second = word.substr(pos + 1);
}
bpe_ranks.emplace(std::make_pair(first, second), i);
}
}
// default special tokens
@@ -2226,6 +2233,7 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
|| t.first == "<|end_of_text|>" // granite
|| t.first == "<EOT>"
|| t.first == "_<EOT>"
|| t.first == "[EOT]" // Kimi-K2
|| t.first == "<｜end▁of▁sentence｜>" // DeepSeek
|| t.first == "<end_of_utterance>" // smoldocling
) {
@@ -2322,6 +2330,7 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
|| t.first == "<fim-pad>"
|| t.first == "<fim_pad>" // Granite
|| t.first == "<PAD>"
|| t.first == "[PAD]" // Kimi-K2
) {
special_fim_pad_id = t.second;
if ((attr & LLAMA_TOKEN_ATTR_CONTROL) == 0) {
@@ -2424,6 +2433,8 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
|| t.first == "<|eom_id|>"
|| t.first == "<EOT>"
|| t.first == "_<EOT>"
|| t.first == "[EOT]" // Kimi-K2
|| t.first == "[EOS]" // Kimi-K2
|| t.first == "<|end_of_text|>"
|| t.first == "<end_of_utterance>" // smoldocling
) {