From 41678c8d5b6ebb3d18009696a0f999228375def6 Mon Sep 17 00:00:00 2001 From: Xuan Son Nguyen Date: Mon, 1 Jul 2024 10:38:52 -0700 Subject: [PATCH] Add sliding window mask for Gemma2 This is a cherry-pick of ggerganov/llama.cpp#8227 --- llama.cpp/llama.cpp | 102 +++++++++++++++++++++++++++++++------------- 1 file changed, 72 insertions(+), 30 deletions(-) diff --git a/llama.cpp/llama.cpp b/llama.cpp/llama.cpp index ffb08ff80..3ee82ca1e 100644 --- a/llama.cpp/llama.cpp +++ b/llama.cpp/llama.cpp @@ -256,6 +256,7 @@ enum llm_kv { LLM_KV_ATTENTION_LAYERNORM_EPS, LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, LLM_KV_ATTENTION_CAUSAL, + LLM_KV_ATTENTION_SLIDING_WINDOW, LLM_KV_ROPE_DIMENSION_COUNT, LLM_KV_ROPE_FREQ_BASE, @@ -336,6 +337,7 @@ static const std::map LLM_KV_NAMES = { { LLM_KV_ATTENTION_LAYERNORM_EPS, "%s.attention.layer_norm_epsilon" }, { LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, "%s.attention.layer_norm_rms_epsilon" }, { LLM_KV_ATTENTION_CAUSAL, "%s.attention.causal" }, + { LLM_KV_ATTENTION_SLIDING_WINDOW, "%s.attention.sliding_window" }, { LLM_KV_ROPE_DIMENSION_COUNT, "%s.rope.dimension_count" }, { LLM_KV_ROPE_FREQ_BASE, "%s.rope.freq_base" }, @@ -1519,6 +1521,7 @@ struct llama_hparams { uint32_t n_head_kv; uint32_t n_layer; uint32_t n_rot; + uint32_t n_swa = 0; // sliding window attention (SWA) uint32_t n_embd_head_k; // dimension of keys (d_k). d_q is assumed to be the same, but there are n_head q heads, and only n_head_kv k-v heads uint32_t n_embd_head_v; // dimension of values (d_v) aka n_embd_head uint32_t n_ff; @@ -1564,6 +1567,7 @@ struct llama_hparams { if (this->n_head_kv != other.n_head_kv) return true; if (this->n_layer != other.n_layer) return true; if (this->n_rot != other.n_rot) return true; + if (this->n_swa != other.n_swa) return true; if (this->n_embd_head_k != other.n_embd_head_k) return true; if (this->n_embd_head_v != other.n_embd_head_v) return true; if (this->n_ff != other.n_ff) return true; @@ -2026,17 +2030,18 @@ struct llama_context { void * abort_callback_data = nullptr; // input tensors - struct ggml_tensor * inp_tokens; // I32 [n_batch] - struct ggml_tensor * inp_embd; // F32 [n_embd, n_batch] - struct ggml_tensor * inp_pos; // I32 [n_batch] - struct ggml_tensor * inp_out_ids; // I32 [n_outputs] - struct ggml_tensor * inp_KQ_mask; // F32 [kv_size, n_batch] - struct ggml_tensor * inp_K_shift; // I32 [kv_size] - struct ggml_tensor * inp_mean; // F32 [n_batch, n_batch] - struct ggml_tensor * inp_cls; // I32 [n_batch] - struct ggml_tensor * inp_s_copy; // I32 [kv_size] - struct ggml_tensor * inp_s_mask; // F32 [1, n_kv] - struct ggml_tensor * inp_s_seq; // I32 [n_kv, n_batch] + struct ggml_tensor * inp_tokens; // I32 [n_batch] + struct ggml_tensor * inp_embd; // F32 [n_embd, n_batch] + struct ggml_tensor * inp_pos; // I32 [n_batch] + struct ggml_tensor * inp_out_ids; // I32 [n_outputs] + struct ggml_tensor * inp_KQ_mask; // F32 [kv_size, n_batch] + struct ggml_tensor * inp_KQ_mask_swa; // F32 [kv_size, n_batch] + struct ggml_tensor * inp_K_shift; // I32 [kv_size] + struct ggml_tensor * inp_mean; // F32 [n_batch, n_batch] + struct ggml_tensor * inp_cls; // I32 [n_batch] + struct ggml_tensor * inp_s_copy; // I32 [kv_size] + struct ggml_tensor * inp_s_mask; // F32 [1, n_kv] + struct ggml_tensor * inp_s_seq; // I32 [n_kv, n_batch] // control vectors struct llama_control_vector cvec; @@ -4002,6 +4007,8 @@ static void llm_load_hparams( } break; case LLM_ARCH_GEMMA2: { + hparams.n_swa = 4096; // default value of gemma 2 + ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa, false); ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps); ml.get_key(LLM_KV_ATTN_LOGIT_SOFTCAPPING, hparams.f_attn_logit_softcapping, false); ml.get_key(LLM_KV_FINAL_LOGIT_SOFTCAPPING, hparams.f_final_logit_softcapping, false); @@ -4543,6 +4550,7 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) { LLAMA_LOG_INFO("%s: n_head_kv = %u\n", __func__, hparams.n_head_kv); LLAMA_LOG_INFO("%s: n_layer = %u\n", __func__, hparams.n_layer); LLAMA_LOG_INFO("%s: n_rot = %u\n", __func__, hparams.n_rot); + LLAMA_LOG_INFO("%s: n_swa = %u\n", __func__, hparams.n_swa); LLAMA_LOG_INFO("%s: n_embd_head_k = %u\n", __func__, hparams.n_embd_head_k); LLAMA_LOG_INFO("%s: n_embd_head_v = %u\n", __func__, hparams.n_embd_head_v); LLAMA_LOG_INFO("%s: n_gqa = %u\n", __func__, hparams.n_gqa()); @@ -6704,17 +6712,18 @@ struct llm_build_context { ctx0 = ggml_init(params); - lctx.inp_tokens = nullptr; - lctx.inp_embd = nullptr; - lctx.inp_pos = nullptr; - lctx.inp_out_ids = nullptr; - lctx.inp_KQ_mask = nullptr; - lctx.inp_K_shift = nullptr; - lctx.inp_mean = nullptr; - lctx.inp_cls = nullptr; - lctx.inp_s_copy = nullptr; - lctx.inp_s_mask = nullptr; - lctx.inp_s_seq = nullptr; + lctx.inp_tokens = nullptr; + lctx.inp_embd = nullptr; + lctx.inp_pos = nullptr; + lctx.inp_out_ids = nullptr; + lctx.inp_KQ_mask = nullptr; + lctx.inp_KQ_mask_swa = nullptr; + lctx.inp_K_shift = nullptr; + lctx.inp_mean = nullptr; + lctx.inp_cls = nullptr; + lctx.inp_s_copy = nullptr; + lctx.inp_s_mask = nullptr; + lctx.inp_s_seq = nullptr; } void free() { @@ -6868,16 +6877,26 @@ struct llm_build_context { } struct ggml_tensor * build_inp_KQ_mask(bool causal = true) { - if (causal) { - lctx.inp_KQ_mask = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)); - } else { - lctx.inp_KQ_mask = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_tokens, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)); - } + lctx.inp_KQ_mask = causal + ? ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)) + : ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_tokens, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)); cb(lctx.inp_KQ_mask, "KQ_mask", -1); ggml_set_input(lctx.inp_KQ_mask); return flash_attn ? ggml_cast(ctx0, lctx.inp_KQ_mask, GGML_TYPE_F16) : lctx.inp_KQ_mask; } + struct ggml_tensor * build_inp_KQ_mask_swa(bool causal = true) { + GGML_ASSERT(hparams.n_swa > 0); + + lctx.inp_KQ_mask_swa = causal + ? ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_kv, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)) + : ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_tokens, GGML_PAD(n_tokens, GGML_KQ_MASK_PAD)); + cb(lctx.inp_KQ_mask_swa, "KQ_mask_swa", -1); + ggml_set_input(lctx.inp_KQ_mask_swa); + + return flash_attn ? ggml_cast(ctx0, lctx.inp_KQ_mask_swa, GGML_TYPE_F16) : lctx.inp_KQ_mask_swa; + } + struct ggml_tensor * build_inp_mean() { lctx.inp_mean = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_tokens, n_tokens); cb(lctx.inp_mean, "inp_mean", -1); @@ -9910,9 +9929,14 @@ struct llm_build_context { struct ggml_tensor * inp_pos = build_inp_pos(); // KQ_mask (mask for 1 head, it will be broadcasted to all heads) - struct ggml_tensor * KQ_mask = build_inp_KQ_mask(); + // gemma 2 requires different mask for layers using sliding window (SWA) + struct ggml_tensor * KQ_mask = build_inp_KQ_mask(true); + struct ggml_tensor * KQ_mask_swa = build_inp_KQ_mask_swa(true); for (int il = 0; il < n_layer; ++il) { + // (il % 2) layers use SWA + struct ggml_tensor * KQ_mask_l = (il % 2 == 0) ? KQ_mask_swa : KQ_mask; + // norm cur = llm_build_norm(ctx0, inpL, hparams, model.layers[il].attn_norm, NULL, @@ -9948,9 +9972,14 @@ struct llm_build_context { cur = llm_build_kv(ctx0, model, hparams, cparams, kv_self, gf, model.layers[il].wo, NULL, - Kcur, Vcur, Qcur, KQ_mask, n_tokens, kv_head, n_kv, 1.0f, cb, il); + Kcur, Vcur, Qcur, KQ_mask_l, n_tokens, kv_head, n_kv, 1.0f, cb, il); } + cur = llm_build_norm(ctx0, cur, hparams, + model.layers[il].attn_post_norm, NULL, + LLM_NORM_RMS, cb, il); + cb(cur, "attn_post_norm", il); + if (il == n_layer - 1) { // skip computing output for unused tokens struct ggml_tensor * inp_out_ids = build_inp_out_ids(); @@ -10877,7 +10906,12 @@ static void llama_set_inputs(llama_context & lctx, const llama_batch & batch) { GGML_ASSERT(ggml_backend_buffer_is_host(lctx.inp_KQ_mask->buffer)); - float * data = (float *) lctx.inp_KQ_mask->data; + float * data = (float *) lctx.inp_KQ_mask->data; + float * data_swa = nullptr; + + if (lctx.inp_KQ_mask_swa) { + data_swa = (float *) lctx.inp_KQ_mask_swa->data; + } // For causal attention, use only the previous KV cells // of the correct sequence for each token of the batch. @@ -10899,6 +10933,14 @@ static void llama_set_inputs(llama_context & lctx, const llama_batch & batch) { } } data[h*(n_kv*n_tokens) + j*n_kv + i] = f; + + // may need to cut off old tokens for sliding window + if (data_swa) { + if (pos - lctx.kv_self.cells[i].pos >= (int32_t)hparams.n_swa) { + f = -INFINITY; + } + data_swa[h*(n_kv*n_tokens) + j*n_kv + i] = f; + } } }