Skip to content
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

Llama 3.1: replace for loop by tensor ops at inv_freq initialization #32244

Merged
merged 2 commits into from
Jul 27, 2024
Merged

Llama 3.1: replace for loop by tensor ops at inv_freq initialization #32244

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c5c3f3d
replace for loop by tensor ops
gante Jul 26, 2024
e2b79a8
rm assert; readability
gante Jul 27, 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
25 changes: 12 additions & 13 deletions src/transformers/modeling_rope_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -324,18 +324,17 @@ def _compute_llama3_parameters(

low_freq_wavelen = old_context_len / low_freq_factor
high_freq_wavelen = old_context_len / high_freq_factor
new_freqs = []
for freq in inv_freq:
wavelen = 2 * math.pi / freq
if wavelen < high_freq_wavelen:
new_freqs.append(freq)
elif wavelen > low_freq_wavelen:
new_freqs.append(freq / factor)
else:
assert low_freq_wavelen != high_freq_wavelen
smooth = (old_context_len / wavelen - low_freq_factor) / (high_freq_factor - low_freq_factor)
new_freqs.append((1 - smooth) * freq / factor + smooth * freq)
inv_freq = torch.tensor(new_freqs, dtype=inv_freq.dtype, device=inv_freq.device)

wavelen = 2 * math.pi / inv_freq
# wavelen < high_freq_wavelen: do nothing
# wavelen > low_freq_wavelen: divide by factor
inv_freq_new = torch.where(wavelen > low_freq_wavelen, inv_freq / factor, inv_freq)
# otherwise: interpolate between the two, using a smooth factor
smooth_factor = (old_context_len / wavelen - low_freq_factor) / (high_freq_factor - low_freq_factor)
smoothed_inv_freq = (1 - smooth_factor) * inv_freq_new / factor + smooth_factor * inv_freq_new
is_medium_freq = ~(wavelen < high_freq_wavelen) * ~(wavelen > low_freq_wavelen)
inv_freq_new = torch.where(is_medium_freq, smoothed_inv_freq, inv_freq_new)

return inv_freq, attention_factor


Expand Down Expand Up @@ -501,7 +500,7 @@ def _validate_llama3_parameters(config: PretrainedConfig):
logger.warning(f"`rope_scaling`'s low_freq_factor field must be a float, got {low_freq_factor}")
if high_freq_factor is None or not isinstance(high_freq_factor, float):
logger.warning(f"`rope_scaling`'s high_freq_factor field must be a float, got {high_freq_factor}")
if high_freq_factor < low_freq_factor:
if high_freq_factor <= low_freq_factor:
logger.warning(
"`rope_scaling`'s high_freq_factor field must be greater than low_freq_factor, got high_freq_factor="
f"{high_freq_factor} and low_freq_factor={low_freq_factor}"
Expand Down
Loading