Skip to content

llama : fix llm_build_k_shift to use correct n_rot #4889

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jan 12, 2024
Merged

llama : fix llm_build_k_shift to use correct n_rot #4889

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ff0899c
llama : fix llm_build_k_shift to use correct n_rot
ggerganov Jan 12, 2024
0cb764e
llama : always use hparams.n_rot for ggml_rope_custom
ggerganov Jan 12, 2024
eea1903
convert : fix persimmon conversion to write correct n_rot
ggerganov Jan 12, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions common/common.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1055,6 +1055,9 @@ struct llama_model_params llama_model_params_from_gpt_params(const gpt_params &
}

static ggml_type kv_cache_type_from_str(const std::string & s) {
if (s == "f32") {
return GGML_TYPE_F32;
}
if (s == "f16") {
return GGML_TYPE_F16;
}
Expand Down
9 changes: 8 additions & 1 deletion convert-hf-to-gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -817,10 +817,17 @@ def set_gguf_parameters(self):
hidden_size = self.hparams["hidden_size"]

self.gguf_writer.add_name('persimmon-8b-chat')
self.gguf_writer.add_context_length(self.hparams["max_position_embeddings"])
self.gguf_writer.add_embedding_length(hidden_size)
self.gguf_writer.add_block_count(block_count)
self.gguf_writer.add_feed_forward_length(self.hparams["intermediate_size"])
self.gguf_writer.add_rope_dimension_count(hidden_size // head_count)

# NOTE: not sure about this change - why does the model not have a rope dimension count when it is smaller
# than the head size?
# ref: https://github.com/ggerganov/llama.cpp/pull/4889
#self.gguf_writer.add_rope_dimension_count(hidden_size // head_count)
self.gguf_writer.add_rope_dimension_count(hidden_size // head_count // 2)

self.gguf_writer.add_head_count(head_count)
self.gguf_writer.add_head_count_kv(head_count_kv)
self.gguf_writer.add_rope_freq_base(self.hparams["rope_theta"])
Expand Down
7 changes: 7 additions & 0 deletions gguf-py/gguf/tensor_mapping.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,6 +57,7 @@ class TensorNameMap:
"transformer.norm_f", # mpt
"ln_f", # refact bloom qwen gpt2
"language_model.encoder.final_layernorm", # persimmon
"model.final_layernorm", # persimmon
"lm_head.ln", # phi2
),

Expand Down Expand Up @@ -98,6 +99,7 @@ class TensorNameMap:
"transformer.h.{bid}.self_attention.query_key_value", # falcon
"h.{bid}.self_attention.query_key_value", # bloom
"language_model.encoder.layers.{bid}.self_attention.query_key_value", # persimmon
"model.layers.{bid}.self_attn.query_key_value", # persimmon
"h.{bid}.attn.c_attn", # gpt2
"transformer.h.{bid}.mixer.Wqkv", # phi2
),
Expand Down Expand Up @@ -141,6 +143,7 @@ class TensorNameMap:
"encoder.layer.{bid}.attention.output.dense", # bert
"transformer.h.{bid}.attn.out_proj", # gpt-j
"language_model.encoder.layers.{bid}.self_attention.dense", # persimmon
"model.layers.{bid}.self_attn.dense", # persimmon
"h.{bid}.attn.c_proj", # gpt2
"transformer.h.{bid}.mixer.out_proj", # phi2
"model.layers.layers.{bid}.self_attn.o_proj", # plamo
Expand Down Expand Up @@ -184,6 +187,7 @@ class TensorNameMap:
"encoder.layer.{bid}.intermediate.dense", # bert
"transformer.h.{bid}.mlp.fc_in", # gpt-j
"language_model.encoder.layers.{bid}.mlp.dense_h_to_4h", # persimmon
"model.layers.{bid}.mlp.dense_h_to_4h", # persimmon
"transformer.h.{bid}.mlp.w1", # qwen
"h.{bid}.mlp.c_fc", # gpt2
"transformer.h.{bid}.mlp.fc1", # phi2
Expand Down Expand Up @@ -225,6 +229,7 @@ class TensorNameMap:
"encoder.layer.{bid}.output.dense", # bert
"transformer.h.{bid}.mlp.fc_out", # gpt-j
"language_model.encoder.layers.{bid}.mlp.dense_4h_to_h", # persimmon
"model.layers.{bid}.mlp.dense_4h_to_h", # persimmon
"h.{bid}.mlp.c_proj", # gpt2
"transformer.h.{bid}.mlp.fc2", # phi2
"model.layers.layers.{bid}.mlp.down_proj", # plamo
Expand All @@ -237,10 +242,12 @@ class TensorNameMap:

MODEL_TENSOR.ATTN_Q_NORM: (
"language_model.encoder.layers.{bid}.self_attention.q_layernorm",
"model.layers.{bid}.self_attn.q_layernorm", # persimmon
),

MODEL_TENSOR.ATTN_K_NORM: (
"language_model.encoder.layers.{bid}.self_attention.k_layernorm",
"model.layers.{bid}.self_attn.k_layernorm", # persimmon
),

MODEL_TENSOR.ROPE_FREQS: (
Expand Down
65 changes: 33 additions & 32 deletions llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4104,22 +4104,20 @@ static void llm_build_k_shift(
struct ggml_cgraph * graph,
llm_rope_type type,
int64_t n_ctx,
int n_rot,
float freq_base,
float freq_scale,
const llm_build_cb & cb) {
const int64_t n_layer = hparams.n_layer;
const int64_t n_head_kv = hparams.n_head_kv;
const int64_t n_embd_head_k = hparams.n_embd_head_k;
const int64_t n_embd_k_gqa = hparams.n_embd_k_gqa();
const int32_t n_rot = hparams.n_rot;
const int32_t n_orig_ctx = cparams.n_yarn_orig_ctx;
const float ext_factor = cparams.yarn_ext_factor;
const float attn_factor = cparams.yarn_attn_factor;
const float beta_fast = cparams.yarn_beta_fast;
const float beta_slow = cparams.yarn_beta_slow;

GGML_ASSERT(n_embd_head_k % n_rot == 0);

struct ggml_tensor * K_shift = ggml_new_tensor_1d(ctx, GGML_TYPE_I32, n_ctx);
cb(K_shift, "K_shift", -1);

Expand Down Expand Up @@ -4523,7 +4521,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -4561,14 +4559,14 @@ struct llm_build_context {

Qcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens), inp_pos,
n_embd_head, 0, 0, n_orig_ctx, freq_base, freq_scale,
hparams.n_rot, 0, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Qcur, "Qcur", il);

Kcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens), inp_pos,
n_embd_head, 0, 0, n_orig_ctx, freq_base, freq_scale,
hparams.n_rot, 0, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Kcur, "Kcur", il);
Expand Down Expand Up @@ -4691,6 +4689,7 @@ struct llm_build_context {

const int64_t n_embd_head = hparams.n_embd_head_v;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head == hparams.n_rot);

struct ggml_tensor * cur;
struct ggml_tensor * inpL;
Expand All @@ -4708,7 +4707,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand All @@ -4734,12 +4733,12 @@ struct llm_build_context {
case MODEL_7B:
Qcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens), inp_pos,
n_embd_head, 0, 0, n_orig_ctx, freq_base, freq_scale,
hparams.n_rot, 0, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens), inp_pos,
n_embd_head, 0, 0, n_orig_ctx, freq_base, freq_scale,
hparams.n_rot, 0, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
break;
Expand Down Expand Up @@ -4812,6 +4811,7 @@ struct llm_build_context {
const int64_t n_embd_head = hparams.n_embd_head_v;
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head == hparams.n_rot);

struct ggml_tensor * cur;
struct ggml_tensor * inpL;
Expand All @@ -4829,7 +4829,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -4870,13 +4870,13 @@ struct llm_build_context {

// using mode = 2 for neox mode
Qcur = ggml_rope_custom(
ctx0, Qcur, inp_pos, n_embd_head, 2, 0, n_orig_ctx,
ctx0, Qcur, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Qcur, "Qcur", il);

Kcur = ggml_rope_custom(
ctx0, Kcur, inp_pos, n_embd_head, 2, 0, n_orig_ctx,
ctx0, Kcur, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Kcur, "Kcur", il);
Expand Down Expand Up @@ -5033,9 +5033,8 @@ struct llm_build_context {
struct ggml_cgraph * gf = ggml_new_graph_custom(ctx0, LLAMA_MAX_NODES, false);

const int64_t n_embd_head = hparams.n_embd_head_v;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);

const int64_t n_rot = n_embd_head_k / 2;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head/2 == hparams.n_rot);

struct ggml_tensor * cur;
struct ggml_tensor * inpL;
Expand All @@ -5052,7 +5051,7 @@ struct llm_build_context {
cb(KQ_mask, "KQ_mask", -1);

if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -5112,15 +5111,15 @@ struct llm_build_context {

// RoPE the first n_rot of q/k, pass the other half, and concat.
struct ggml_tensor * qrot = ggml_view_3d(
ctx0, tmpq, n_rot, n_head, n_tokens,
ctx0, tmpq, hparams.n_rot, n_head, n_tokens,
ggml_element_size(tmpq) * n_embd_head,
ggml_element_size(tmpq) * n_embd_head * n_head,
0
);
cb(qrot, "qrot", il);

struct ggml_tensor * krot = ggml_view_3d(
ctx0, tmpk, n_rot, n_head, n_tokens,
ctx0, tmpk, hparams.n_rot, n_head, n_tokens,
ggml_element_size(tmpk) * n_embd_head,
ggml_element_size(tmpk) * n_embd_head * n_head,
0
Expand All @@ -5129,29 +5128,29 @@ struct llm_build_context {

// get the second half of tmpq, e.g tmpq[n_rot:, :, :]
struct ggml_tensor * qpass = ggml_view_3d(
ctx0, tmpq, n_rot, n_head, n_tokens,
ctx0, tmpq, hparams.n_rot, n_head, n_tokens,
ggml_element_size(tmpq) * n_embd_head,
ggml_element_size(tmpq) * n_embd_head * n_head,
ggml_element_size(tmpq) * n_rot
ggml_element_size(tmpq) * hparams.n_rot
);
cb(qpass, "qpass", il);

struct ggml_tensor * kpass = ggml_view_3d(
ctx0, tmpk, n_rot, n_head, n_tokens,
ctx0, tmpk, hparams.n_rot, n_head, n_tokens,
ggml_element_size(tmpk) * n_embd_head,
ggml_element_size(tmpk) * n_embd_head * n_head,
ggml_element_size(tmpk) * n_rot
ggml_element_size(tmpk) * hparams.n_rot
);
cb(kpass, "kpass", il);

struct ggml_tensor * qrotated = ggml_rope_custom(
ctx0, qrot, inp_pos, n_rot, 2, 0, n_orig_ctx,
ctx0, qrot, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(qrotated, "qrotated", il);

struct ggml_tensor * krotated = ggml_rope_custom(
ctx0, krot, inp_pos, n_rot, 2, 0, n_orig_ctx,
ctx0, krot, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(krotated, "krotated", il);
Expand Down Expand Up @@ -5531,6 +5530,7 @@ struct llm_build_context {

const int64_t n_embd_head = hparams.n_embd_head_v;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head == hparams.n_rot);

struct ggml_tensor * cur;
struct ggml_tensor * inpL;
Expand All @@ -5548,7 +5548,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, hparams.n_rot, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -5661,7 +5661,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -5693,13 +5693,13 @@ struct llm_build_context {

// using mode = 2 for neox mode
Qcur = ggml_rope_custom(
ctx0, Qcur, inp_pos, n_embd_head, 2, 0, n_orig_ctx,
ctx0, Qcur, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Qcur, "Qcur", il);

Kcur = ggml_rope_custom(
ctx0, Kcur, inp_pos, n_embd_head, 2, 0, n_orig_ctx,
ctx0, Kcur, inp_pos, hparams.n_rot, 2, 0, n_orig_ctx,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Kcur, "Kcur", il);
Expand Down Expand Up @@ -5778,7 +5778,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE_NEOX, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand Down Expand Up @@ -5874,6 +5874,7 @@ struct llm_build_context {

const int64_t n_embd_head = hparams.n_embd_head_v;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head == hparams.n_rot);

struct ggml_tensor * cur;
struct ggml_tensor * inpL;
Expand All @@ -5891,7 +5892,7 @@ struct llm_build_context {

// shift the entire K-cache if needed
if (do_rope_shift) {
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, n_embd_head, freq_base, freq_scale, cb);
llm_build_k_shift(ctx0, hparams, cparams, kv_self, gf, LLM_ROPE, n_ctx, freq_base, freq_scale, cb);
}

for (int il = 0; il < n_layer; ++il) {
Expand All @@ -5917,13 +5918,13 @@ struct llm_build_context {
cb(Vcur, "Vcur", il);

Qcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens), inp_pos,
ctx0, ggml_reshape_3d(ctx0, Qcur, hparams.n_rot, n_head, n_tokens), inp_pos,
n_embd_head, 2, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Qcur, "Qcur", il);

Kcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens), inp_pos,
ctx0, ggml_reshape_3d(ctx0, Kcur, hparams.n_rot, n_head_kv, n_tokens), inp_pos,
n_embd_head, 2, 0, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Kcur, "Kcur", il);
Expand Down