vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama: add support for QRWKV6 model architecture (#11001)

Browse source 
llama: add support for QRWKV6 model architecture (#11001)

* WIP: Add support for RWKV6Qwen2

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* RWKV: Some graph simplification

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Add support for RWKV6Qwen2 with cpu and cuda GLA

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* RWKV6[QWEN2]: Concat lerp weights together to reduce cpu overhead

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Fix some typos

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* code format changes

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Fix wkv test & add gla test

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Fix cuda warning

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Update README.md

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* Update ggml/src/ggml-cuda/gla.cu

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* Fix fused lerp weights loading with RWKV6

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>

* better sanity check skipping for QRWKV6 in llama-quant

thanks @compilade

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>
Co-authored-by: compilade <git@compilade.net>

---------

Signed-off-by: Molly Sophia <mollysophia379@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: compilade <git@compilade.net>
This commit is contained in:
Molly Sophia 2025-01-10 09:58:08 +08:00 • committed by GitHub
parent c6860cc734
commit ee7136c6d1
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
23 changed files with 862 additions and 124 deletions
Download patch file Download diff file Expand all files Collapse all files

30
src/llama-arch.cpp
View file

@ -57,6 +57,7 @@ static const std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
{ LLM_ARCH_NEMOTRON, "nemotron" },
{ LLM_ARCH_EXAONE, "exaone" },
{ LLM_ARCH_RWKV6, "rwkv6" },
{ LLM_ARCH_RWKV6QWEN2, "rwkv6qwen2" },
{ LLM_ARCH_GRANITE, "granite" },
{ LLM_ARCH_GRANITE_MOE, "granitemoe" },
{ LLM_ARCH_CHAMELEON, "chameleon" },
@ -106,6 +107,7 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
{ LLM_KV_TIME_DECAY_EXTRA_DIM, "%s.time_decay_extra_dim" },
{ LLM_KV_RESIDUAL_SCALE, "%s.residual_scale" },
{ LLM_KV_EMBEDDING_SCALE, "%s.embedding_scale" },
{ LLM_KV_TOKEN_SHIFT_COUNT, "%s.token_shift_count" },
{ LLM_KV_ATTENTION_HEAD_COUNT, "%s.attention.head_count" },
{ LLM_KV_ATTENTION_HEAD_COUNT_KV, "%s.attention.head_count_kv" },
@ -1166,6 +1168,7 @@ static const std::map<llm_arch, std::map<llm_tensor, const char *>> LLM_TENSOR_N
{ LLM_TENSOR_TIME_MIX_LERP_V, "blk.%d.time_mix_lerp_v" },
{ LLM_TENSOR_TIME_MIX_LERP_R, "blk.%d.time_mix_lerp_r" },
{ LLM_TENSOR_TIME_MIX_LERP_G, "blk.%d.time_mix_lerp_g" },
{ LLM_TENSOR_TIME_MIX_LERP_FUSED, "blk.%d.time_mix_lerp_fused" },
{ LLM_TENSOR_TIME_MIX_FIRST, "blk.%d.time_mix_first" },
{ LLM_TENSOR_TIME_MIX_DECAY, "blk.%d.time_mix_decay" },
{ LLM_TENSOR_TIME_MIX_DECAY_W1, "blk.%d.time_mix_decay_w1" },
@ -1183,6 +1186,32 @@ static const std::map<llm_arch, std::map<llm_tensor, const char *>> LLM_TENSOR_N
{ LLM_TENSOR_CHANNEL_MIX_RECEPTANCE, "blk.%d.channel_mix_receptance" },
},
},
{
LLM_ARCH_RWKV6QWEN2,
{
{ LLM_TENSOR_TOKEN_EMBD, "token_embd" },
{ LLM_TENSOR_OUTPUT_NORM, "output_norm" },
{ LLM_TENSOR_OUTPUT, "output" },
{ LLM_TENSOR_ATTN_NORM, "blk.%d.attn_norm" },
{ LLM_TENSOR_TIME_MIX_W1, "blk.%d.time_mix_w1" },
{ LLM_TENSOR_TIME_MIX_W2, "blk.%d.time_mix_w2" },
{ LLM_TENSOR_TIME_MIX_LERP_X, "blk.%d.time_mix_lerp_x" },
{ LLM_TENSOR_TIME_MIX_LERP_FUSED, "blk.%d.time_mix_lerp_fused" },
{ LLM_TENSOR_TIME_MIX_FIRST, "blk.%d.time_mix_first" },
{ LLM_TENSOR_TIME_MIX_DECAY, "blk.%d.time_mix_decay" },
{ LLM_TENSOR_TIME_MIX_DECAY_W1, "blk.%d.time_mix_decay_w1" },
{ LLM_TENSOR_TIME_MIX_DECAY_W2, "blk.%d.time_mix_decay_w2" },
{ LLM_TENSOR_TIME_MIX_KEY, "blk.%d.time_mix_key" },
{ LLM_TENSOR_TIME_MIX_VALUE, "blk.%d.time_mix_value" },
{ LLM_TENSOR_TIME_MIX_RECEPTANCE, "blk.%d.time_mix_receptance" },
{ LLM_TENSOR_TIME_MIX_GATE, "blk.%d.time_mix_gate" },
{ LLM_TENSOR_TIME_MIX_OUTPUT, "blk.%d.time_mix_output" },
{ LLM_TENSOR_FFN_NORM, "blk.%d.ffn_norm" },
{ LLM_TENSOR_FFN_GATE, "blk.%d.ffn_gate" },
{ LLM_TENSOR_FFN_DOWN, "blk.%d.ffn_down" },
{ LLM_TENSOR_FFN_UP, "blk.%d.ffn_up" },
},
},
{
LLM_ARCH_GRANITE,
{
@ -1365,6 +1394,7 @@ static const std::map<llm_tensor, llm_tensor_info> LLM_TENSOR_INFOS = {
{LLM_TENSOR_TIME_MIX_LERP_V, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ADD}},
{LLM_TENSOR_TIME_MIX_LERP_R, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ADD}},
{LLM_TENSOR_TIME_MIX_LERP_G, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ADD}},
{LLM_TENSOR_TIME_MIX_LERP_FUSED, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ADD}},
{LLM_TENSOR_TIME_MIX_DECAY, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ADD}},
{LLM_TENSOR_TIME_MIX_FIRST, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_RWKV_WKV6}},
{LLM_TENSOR_ATTN_NORM, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL}},