Skip to content

ggml : unify rope norm/neox #7634

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 9 commits into from
Jun 5, 2024
Merged

ggml : unify rope norm/neox #7634

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
2fd31fe
ggml : unify rope norm/neox (CPU)
ggerganov May 30, 2024
e2370c8
ggml : fix compile warning
ggerganov May 30, 2024
cbe4f5f
ggml : remove GLM rope mode
ggerganov May 30, 2024
3035c2d
metal : better rope implementation
ggerganov May 30, 2024
c989fd0
cuda : better rope implementation
ggerganov May 30, 2024
572446b
naming : n_orig_ctx -> n_ctx_orig
ggerganov May 30, 2024
437d2d6
dev : add reminders to update backends
ggerganov May 30, 2024
61e0a84
vulkan : fix ggml_rope_ext() usage
ggerganov May 30, 2024
ddac1ef
cuda : fix array size + indents
ggerganov Jun 2, 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
12 changes: 6 additions & 6 deletions examples/baby-llama/baby-llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -522,8 +522,8 @@ static struct ggml_tensor * forward(
// wk shape [n_embd, n_embd, 1, 1]
// Qcur shape [n_embd/n_head, n_head, N, 1]
// Kcur shape [n_embd/n_head, n_head, N, 1]
struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0, 0);
struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0, 0);
struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0);
struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N), KQ_pos, n_rot, 0);

// store key and value to memory
{
Expand Down Expand Up @@ -759,8 +759,8 @@ static struct ggml_tensor * forward_batch(
// wk shape [n_embd, n_embd, 1, 1]
// Qcur shape [n_embd/n_head, n_head, N, n_batch]
// Kcur shape [n_embd/n_head, n_head, N, n_batch]
struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0, 0);
struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0, 0);
struct ggml_tensor * Qcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wq, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0);
struct ggml_tensor * Kcur = ggml_rope(ctx0, ggml_reshape_4d(ctx0, ggml_mul_mat(ctx0, model->layers[il].wk, cur), n_embd/n_head, n_head, N, n_batch), KQ_pos, n_rot, 0);
assert_shape_4d(Qcur, n_embd/n_head, n_head, N, n_batch);
assert_shape_4d(Kcur, n_embd/n_head, n_head, N, n_batch);

Expand Down Expand Up @@ -1056,7 +1056,7 @@ static struct ggml_tensor * forward_lora(
model->layers[il].wqb,
cur)),
n_embd/n_head, n_head, N),
KQ_pos, n_rot, 0, 0);
KQ_pos, n_rot, 0);
struct ggml_tensor * Kcur = ggml_rope(ctx0,
ggml_reshape_3d(ctx0,
ggml_mul_mat(ctx0,
Expand All @@ -1065,7 +1065,7 @@ static struct ggml_tensor * forward_lora(
model->layers[il].wkb,
cur)),
n_embd/n_head, n_head, N),
KQ_pos, n_rot, 0, 0);
KQ_pos, n_rot, 0);

// store key and value to memory
{
Expand Down
12 changes: 6 additions & 6 deletions examples/convert-legacy-llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -176,7 +176,7 @@ class Params:
rope_scaling_type: gguf.RopeScalingType | None = None
f_rope_freq_base: float | None = None
f_rope_scale: float | None = None
n_orig_ctx: int | None = None
n_ctx_orig: int | None = None
rope_finetuned: bool | None = None

ftype: GGMLFileType | None = None
Expand Down Expand Up @@ -226,7 +226,7 @@ def loadHFTransformerJson(model: LazyModel, config_path: Path) -> Params:
with open(config_path) as f:
config = json.load(f)

rope_scaling_type = f_rope_scale = n_orig_ctx = rope_finetuned = None
rope_scaling_type = f_rope_scale = n_ctx_orig = rope_finetuned = None
rope_scaling = config.get("rope_scaling")

if rope_scaling is not None and (typ := rope_scaling.get("type")):
Expand All @@ -236,7 +236,7 @@ def loadHFTransformerJson(model: LazyModel, config_path: Path) -> Params:
rope_scaling_type = gguf.RopeScalingType.LINEAR
elif typ == "yarn":
rope_scaling_type = gguf.RopeScalingType.YARN
n_orig_ctx = rope_scaling['original_max_position_embeddings']
n_ctx_orig = rope_scaling['original_max_position_embeddings']
rope_finetuned = rope_scaling['finetuned']
else:
raise NotImplementedError(f'Unknown rope scaling type: {typ}')
Expand Down Expand Up @@ -272,7 +272,7 @@ def loadHFTransformerJson(model: LazyModel, config_path: Path) -> Params:
f_rope_freq_base = config.get("rope_theta"),
rope_scaling_type = rope_scaling_type,
f_rope_scale = f_rope_scale,
n_orig_ctx = n_orig_ctx,
n_ctx_orig = n_ctx_orig,
rope_finetuned = rope_finetuned,
)

Expand Down Expand Up @@ -864,8 +864,8 @@ def add_meta_arch(self, params: Params) -> None:
self.gguf.add_rope_scaling_type(params.rope_scaling_type)
self.gguf.add_rope_scaling_factor(params.f_rope_scale)

if params.n_orig_ctx is not None:
self.gguf.add_rope_scaling_orig_ctx_len(params.n_orig_ctx)
if params.n_ctx_orig is not None:
self.gguf.add_rope_scaling_orig_ctx_len(params.n_ctx_orig)

if params.rope_finetuned is not None:
self.gguf.add_rope_scaling_finetuned(params.rope_finetuned)
Expand Down
2 changes: 1 addition & 1 deletion examples/finetune/finetune.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -564,7 +564,7 @@ static struct ggml_tensor * llama_build_lora_finetune_graphs(
const int rope_mode = 0;

return ggml_rope_ext(ctx,
t, KQ_pos, nullptr, n_rot, rope_mode, n_ctx, 0,
t, KQ_pos, nullptr, n_rot, rope_mode, n_ctx,
rope_freq_base, rope_freq_scale, 0.0f, 1.0f, 0.0f, 0.0f
);
};
Expand Down
2 changes: 1 addition & 1 deletion examples/train-text-from-scratch/train-text-from-scratch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -302,7 +302,7 @@ static struct ggml_tensor * llama_build_train_graphs(
const int rope_mode = 0;

return ggml_rope_ext(
ctx, t, KQ_pos, nullptr, n_rot, rope_mode, n_ctx, 0, rope_freq_base, rope_freq_scale, 0.0f, 1.0f, 0.0f, 0.0f
ctx, t, KQ_pos, nullptr, n_rot, rope_mode, n_ctx, rope_freq_base, rope_freq_scale, 0.0f, 1.0f, 0.0f, 0.0f
);
};

Expand Down
Loading
Loading