Skip to content

task: measured text-only slopes for gemma4:31b and qwen3.8 (MLX prefill transient) - #284

Merged
glennneuber merged 1 commit into
mainfrom
docs/mlx-transient-slopes
Sep 7, 2026
Merged

glennneuber merged 1 commit into
mainfrom
docs/mlx-transient-slopes

Conversation

@glennneuber

Copy link
Copy Markdown

Fills the slopes.jsonl placeholder in tasks/mlx-prefill-transient-scales-with-context.md with tonight's measurement (probe-slopes.sh, main-a523d60b, one fresh text prompt per request at num_ctx 65536, beside production) and commits the three probe records under vision-suite/preflight/runs/ (force-added, like the calibration data).

model fit over the ≥ 10k points at a full 65536 #276 prices
gemma4:12b-nvfp4 17.2 + 0.28 GiB per 1k tokens ≈ 36 GiB 22.9 GiB
gemma4:31b-nvfp4 31.3 + 0.58 GiB per 1k ≈ 69 GiB 37.6 GiB
qwen3.8:27b-nvfp4 27.3 + 0.33 GiB per 1k ≈ 49 GiB 31.6 GiB
  • The intercept minus the weights is ADR 0034's calibrated constant, as the amendment says.
  • Within an architecture the slope tracks the weights (gemma4 ≈ 0.035, qwen3.5 dense ≈ 0.019 GiB per 1k tokens per GiB of weights), so fix 1 can be a_arch + b_arch × weights × num_ctx rather than a per-model table. Fix 1 itself is not in this PR.
  • Acceptance criterion 1 is half ticked: measured, formula not yet written.

🤖 Generated with Claude Code

… data in-repo

Text-only prefill peaks at 5.2k/10.4k/20.9k/41.6k tokens on main-a523d60b:
31b 31.3 + 0.58 GiB per 1k tokens (69 GiB at a full 65536 against the
37.6 #276 prices), qwen3.8 27.3 + 0.33 (49 vs 31.6), 12b 17.2 + 0.28
(36 vs 22.9). The intercepts minus weights are ADR 0034's calibrated
constants; within an architecture the slope tracks the weights, so fix 1
can be one constant per architecture times the weights. Criterion 1 is
half ticked (measured; the formula is not written yet). The three probe
records (session growth, growth shape, slopes) are force-added under
preflight/runs like the calibration data.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@glennneuber
glennneuber merged commit 778e015 into main Sep 7, 2026
1 check passed
@glennneuber
glennneuber deleted the docs/mlx-transient-slopes branch September 7, 2026 21:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant