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

Add support for ARWKV7 Hybrid models

Browse source 
Signed-off-by: Molly Sophia <mollysophia379@gmail.com>
This commit is contained in:
Molly Sophia 2025-01-27 17:45:43 +08:00
parent e9ba411d3e
commit f6be4dc661
7 changed files with 420 additions and 72 deletions
Download patch file Download diff file Expand all files Collapse all files

66
src/llama-model.cpp
View file

@ -1211,17 +1211,19 @@ void llama_model::load_hparams(llama_model_loader & ml) {
}
} break;
case LLM_ARCH_RWKV7:
case LLM_ARCH_ARWKV7:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps, false);
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps, false);
ml.get_key(LLM_KV_WKV_HEAD_SIZE, hparams.wkv_head_size);
ml.get_key(LLM_KV_ATTENTION_DECAY_LORA_RANK, hparams.n_lora_decay);
ml.get_key(LLM_KV_ATTENTION_ICLR_LORA_RANK, hparams.n_lora_iclr);
ml.get_key(LLM_KV_ATTENTION_VALUE_RESIDUAL_MIX_LORA_RANK, hparams.n_lora_value_res_mix);
ml.get_key(LLM_KV_ATTENTION_GATE_LORA_RANK, hparams.n_lora_gate);
ml.get_key(LLM_KV_ATTENTION_GATE_LORA_RANK, hparams.n_lora_gate, false);
ml.get_key(LLM_KV_TOKEN_SHIFT_COUNT, hparams.token_shift_count, false);
switch (hparams.n_layer) {
// TODO: Add variants
case 28: type = LLM_TYPE_7B; break; // ARWKV7
default: type = LLM_TYPE_UNKNOWN;
}
} break;
@ -3366,6 +3368,62 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
layer.channel_mix_value = create_tensor(tn(LLM_TENSOR_CHANNEL_MIX_VALUE, "weight", i), {ffn_size, n_embd}, 0);
}
} break;
case LLM_ARCH_ARWKV7:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
// output
output_norm = create_tensor(tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd}, 0);
output = create_tensor(tn(LLM_TENSOR_OUTPUT, "weight"), {n_embd, n_vocab}, 0);
const int n_lora_decay = hparams.n_lora_decay;
const int n_lora_iclr = hparams.n_lora_iclr;
const int n_lora_value_res_mix = hparams.n_lora_value_res_mix;
const int n_lora_gate = hparams.n_lora_gate;
const int attn_hidden_size = n_embd;
for (int i = 0; i < n_layer; ++i) {
auto & layer = layers[i];
layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);
layer.time_mix_w0 = create_tensor(tn(LLM_TENSOR_TIME_MIX_W0, "weight", i), {n_embd}, 0);
layer.time_mix_w1 = create_tensor(tn(LLM_TENSOR_TIME_MIX_W1, "weight", i), {n_embd, n_lora_decay}, 0);
layer.time_mix_w2 = create_tensor(tn(LLM_TENSOR_TIME_MIX_W2, "weight", i), {n_lora_decay, n_embd}, 0);
layer.time_mix_a0 = create_tensor(tn(LLM_TENSOR_TIME_MIX_A0, "weight", i), {n_embd}, 0);
layer.time_mix_a1 = create_tensor(tn(LLM_TENSOR_TIME_MIX_A1, "weight", i), {n_embd, n_lora_iclr}, 0);
layer.time_mix_a2 = create_tensor(tn(LLM_TENSOR_TIME_MIX_A2, "weight", i), {n_lora_iclr, n_embd}, 0);
layer.time_mix_v0 = create_tensor(tn(LLM_TENSOR_TIME_MIX_V0, "weight", i), {n_embd}, 0);
layer.time_mix_v1 = create_tensor(tn(LLM_TENSOR_TIME_MIX_V1, "weight", i), {n_embd, n_lora_value_res_mix}, 0);
layer.time_mix_v2 = create_tensor(tn(LLM_TENSOR_TIME_MIX_V2, "weight", i), {n_lora_value_res_mix, n_embd}, 0);
layer.time_mix_g1 = create_tensor(tn(LLM_TENSOR_TIME_MIX_G1, "weight", i), {n_embd, n_lora_gate}, llama_model_loader::TENSOR_NOT_REQUIRED);
layer.time_mix_g2 = create_tensor(tn(LLM_TENSOR_TIME_MIX_G2, "weight", i), {n_lora_gate, n_embd}, llama_model_loader::TENSOR_NOT_REQUIRED);
layer.time_mix_lerp_fused = create_tensor(tn(LLM_TENSOR_TIME_MIX_LERP_FUSED, "weight", i), {n_embd, 1, 6}, 0);
layer.time_mix_k_k = create_tensor(tn(LLM_TENSOR_TIME_MIX_K_K, "weight", i), {attn_hidden_size}, 0);
layer.time_mix_k_a = create_tensor(tn(LLM_TENSOR_TIME_MIX_K_A, "weight", i), {attn_hidden_size}, 0);
layer.time_mix_r_k = create_tensor(tn(LLM_TENSOR_TIME_MIX_R_K, "weight", i), {attn_hidden_size}, 0);
layer.time_mix_key = create_tensor(tn(LLM_TENSOR_TIME_MIX_KEY, "weight", i), {attn_hidden_size, n_embd}, 0);
layer.time_mix_value = create_tensor(tn(LLM_TENSOR_TIME_MIX_VALUE, "weight", i), {attn_hidden_size, n_embd}, 0);
layer.time_mix_receptance = create_tensor(tn(LLM_TENSOR_TIME_MIX_RECEPTANCE, "weight", i), {attn_hidden_size, n_embd}, 0);
layer.time_mix_ln = create_tensor(tn(LLM_TENSOR_TIME_MIX_LN, "weight", i), {n_embd}, llama_model_loader::TENSOR_NOT_REQUIRED);
layer.time_mix_ln_b = create_tensor(tn(LLM_TENSOR_TIME_MIX_LN, "bias", i), {n_embd}, llama_model_loader::TENSOR_NOT_REQUIRED);
layer.time_mix_output = create_tensor(tn(LLM_TENSOR_TIME_MIX_OUTPUT, "weight", i), {n_embd, attn_hidden_size}, 0);
layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, 0);
layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), { n_ff, n_embd}, 0);
layer.ffn_up = create_tensor(tn(LLM_TENSOR_FFN_UP, "weight", i), {n_embd, n_ff}, 0);
}
} break;
case LLM_ARCH_CHAMELEON:
{
@ -3953,6 +4011,7 @@ enum llama_rope_type llama_model_rope_type(const struct llama_model * model) {
case LLM_ARCH_RWKV6:
case LLM_ARCH_RWKV6QWEN2:
case LLM_ARCH_RWKV7:
case LLM_ARCH_ARWKV7:
case LLM_ARCH_WAVTOKENIZER_DEC:
return LLAMA_ROPE_TYPE_NONE;
@ -4107,6 +4166,7 @@ bool llama_model_is_recurrent(const struct llama_model * model) {
case LLM_ARCH_RWKV6: return true;
case LLM_ARCH_RWKV6QWEN2: return true;
case LLM_ARCH_RWKV7: return true;
case LLM_ARCH_ARWKV7: return true;
default: return false;
}
}