INDEX.md

# SAM 3 research experiments

One row per experiment, newest first. `reconstructed` = rebuilt from history for outputs that predate experiment logging.

| date | experiment | question | status | conclusion |
|---|---|---|---|---|
| 2026-10-15 | [vlm-tiers-gemma-4-31B-it-boxes](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-15_vlm-tiers-gemma-4-31b-it-boxes/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: google/gemma-4-31B-it (35% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-15 | [vlm-tiers-Qwen3.5-9B-boxes](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-15_vlm-tiers-qwen3-5-9b-boxes/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: Qwen/Qwen3.5-9B (29% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-15 | [vlm-tiers-gemma-4-31B-it-v2](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-15_vlm-tiers-gemma-4-31b-it-v2/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: google/gemma-4-31B-it (35% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-15 | [unaccounted-audit-2026-10-14-vlm-tiers-gemma-4-31b-it](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-15_unaccounted-audit-2026-10-14-vlm-tiers-gemma-4-31b-it/summary.md) | Of the frame no tier name covers, how much is mask edges, how much is genuinely unimportant, and how much is something a viewer would care about -- could a fourth "useless" tier take it? | ok | Of the unaccounted 35% of the frame: edges/scraps/'nothing' 30%, judged worth a tier 70%. |
| 2026-10-14 | [stale-prompts-2026-10-12-vlm-tiers-qwen3-5-9b-2](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-14_stale-prompts-2026-10-12-vlm-tiers-qwen3-5-9b-2/summary.md) | If the VLM is asked once and SAM 3 keeps segmenting with those names, how fast does the answer go stale -- and when, on average, should the VLM be asked again? | ok | Median time to trigger within a shot: 1.0 s. |
| 2026-10-14 | [vlm-tiers-InternVL3-5-38B-HF](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-14_vlm-tiers-internvl3-5-38b-hf/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: OpenGVLab/InternVL3_5-38B-HF (40% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-14 | [vlm-tiers-gemma-4-31B-it](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-14_vlm-tiers-gemma-4-31b-it/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: google/gemma-4-31B-it (33% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-12 | [vlm-tiers-Qwen3.5-9B](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-12_vlm-tiers-qwen3-5-9b-2/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | ok | Least unaccounted: Qwen/Qwen3.5-9B (35% of the frame). Check sheets/ for whether the tiers make sense. |
| 2026-10-12 | [vlm-tiers-Qwen3.5-9B](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-12_vlm-tiers-qwen3-5-9b/summary.md) | Can a small open VLM, shown one frame, name everything in it as primary / secondary / tertiary well enough for SAM 3 to turn the names into three tier masks -- and how often must it be asked again? | failed | — |
| 2026-10-12 | [everything-shots](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-12_everything-shots/summary.md) | How much of segment-everything's mask turnover on high-energy footage is shot cuts, how much is motion within a shot, and how much is the segmenter itself? | ok | Noise floor 82%; 1 s within a shot 40%; across a cut 7%; mean shot 8.1 s. Check the cut detector in cuts/. |
| 2026-10-11 | [everything-stress](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-stress/summary.md) | On high-energy footage (sports, gaming, live performance, music), how much does segment-everything cover, how many masks does it make, and how fast does the set of masks turn over from second to second? | ok | High-energy footage: 114 masks/frame covering 82%; 37% of masks persist across 1 s. Per-category rows and sheets/ show where it breaks. |
| 2026-10-11 | [everything-ladder-bfloat16-points-per-side-64](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-points-per-side-64-2/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 64, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 125.6 masks/frame covering 84% (overlap 2.29) in 4.33 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-bfloat16](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-2/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 32, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 77.8 masks/frame covering 81% (overlap 1.96) in 1.15 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-bfloat16-points-per-side-16](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-points-per-side-16-2/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 16, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 42.3 masks/frame covering 78% (overlap 1.62) in 0.34 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-float32](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-float32-2/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'float32', 'points_per_side': 32, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 77.3 masks/frame covering 82% (overlap 1.92) in 1.37 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-bfloat16-crops-1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-crops-1/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | failed | — |
| 2026-10-11 | [everything-ladder-bfloat16-points-per-side-64](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-points-per-side-64/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 64, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 89.2 masks/frame covering 72% (overlap 2.23) in 4.33 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-float32](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-float32/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'float32', 'points_per_side': 32, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 57.3 masks/frame covering 71% (overlap 1.97) in 1.35 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-bfloat16](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 32, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 56.5 masks/frame covering 71% (overlap 1.90) in 1.13 s/frame on NVIDIA H200 NVL. |
| 2026-10-11 | [everything-ladder-bfloat16-points-per-side-16](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-11_everything-ladder-bfloat16-points-per-side-16/summary.md) | How do segment-everything's settings trade speed against coverage and mask granularity, on diverse footage? | ok | {'dtype': 'bfloat16', 'points_per_side': 16, 'crops_n_layers': 0, 'pred_iou': 0.5, 'stability': 0.7}: 31.8 masks/frame covering 65% (overlap 1.58) in 0.33 s/frame on NVIDIA H200 NVL. |
| 2026-10-06 | [hardware-profile-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-06_hardware-profile-native-sam3-1/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA A100-SXM4-40GB (native): load 16 s, warm-up 17 s, then per video setup 3.6 + model 68.1 + render 9.2 s. |
| 2026-10-06 | [hardware-profile-transformers](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-06_hardware-profile-transformers/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA A100-SXM4-40GB (transformers): load 15 s, warm-up 27 s, then per video setup 3.5 + model 82.4 + render 8.3 s. |
| 2026-10-05 | [hardware-profile-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-05_hardware-profile-native-sam3-1/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA H100 PCIe (native): load 15 s, warm-up 7 s, then per video setup 3.3 + model 17.0 + render 11.1 s. |
| 2026-10-05 | [hardware-profile-transformers](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-05_hardware-profile-transformers/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA H100 PCIe (transformers): load 14 s, warm-up 12 s, then per video setup 3.6 + model 20.7 + render 9.9 s. |
| 2026-10-04 | [native-classification](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_native-classification-2/summary.md) | Without being told what to look for, what can SAM 3 find on its own -- which concepts from a fixed vocabulary it detects, how much of the frame "segment everything" covers, and how much of that it can name? | ok | Vocabulary sweep names 17.5 masks/frame (51% of the frame); segment-everything finds 57.3 (71%). See sheets/ for each frame. |
| 2026-10-04 | [native-classification](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_native-classification/summary.md) | Without being told what to look for, what can SAM 3 find on its own -- which concepts from a fixed vocabulary it detects, how much of the frame "segment everything" covers, and how much of that it can name? | failed | — |
| 2026-10-04 | [hardware-profile-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_hardware-profile-native-sam3-1-3/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA L40S (native): load 17 s, warm-up 9 s, then per video setup 3.4 + model 27.6 + render 11.5 s. |
| 2026-10-04 | [hardware-profile-transformers](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_hardware-profile-transformers-2-2/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA L40S (transformers): load 16 s, warm-up 15 s, then per video setup 3.6 + model 22.8 + render 9.8 s. |
| 2026-10-04 | [hardware-profile-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_hardware-profile-native-sam3-1-2/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA H200 NVL (native): load 11 s, warm-up 8 s, then per video setup 1.5 + model 9.3 + render 2.9 s. |
| 2026-10-04 | [hardware-profile-transformers](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_hardware-profile-transformers-3/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | NVIDIA H200 NVL (transformers): load 13 s, warm-up 8 s, then per video setup 1.2 + model 13.0 + render 2.7 s. |
| 2026-10-04 | [hardware-profile-transformers](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_hardware-profile-transformers/summary.md) | How long does each phase of SAM 3 video tracking take on this machine -- load, warm-up, setup, model time, rendering -- and how much does it vary between runs? | ok | Apple M3 Max (transformers): load 4 s, warm-up 71 s, then per video setup 1.6 + model 524.1 + render 2.7 s. |
| 2026-10-04 | [tracking-backend-native-sam3.1-compile-pad-to-16](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-04_tracking-backend-native-sam3-1-compile-pad-to-16/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=flash_attention_3 / compiled: inference 15.6 fps, 1.72x the transformers baseline. 2 ids. |
| 2026-10-03 | [compile-shapes](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_compile-shapes/summary.md) | What makes compiled SAM 3.1 recompile: video length, a leftover post-processing batch, or the way frames are loaded? | ok | Warm-up 219 s. Stalled passes: A same length, C leftover batch. Reasons in the job log (Recompiling ...). |
| 2026-10-03 | [tracking-backend-native-sam3.1-compile-warm-up](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_tracking-backend-native-sam3-1-compile-warm-up/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=flash_attention_3 / compiled: inference 7.5 fps, 0.83x the transformers baseline. 2 ids. |
| 2026-10-03 | [tracking-backend-native-sam3.1-compile-warm-up](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_tracking-backend-native-sam3-1-compile-warm-up-repeat/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=flash_attention_3 / compiled: inference 7.6 fps, 0.84x the transformers baseline. 2 ids. |
| 2026-10-03 | [tracking-backend-native-sam3.1-compile](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_tracking-backend-native-sam3-1-compile/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=flash_attention_3 / compiled: inference 1.0 fps, 0.11x the transformers baseline. 2 ids. |
| 2026-10-03 | [tracking-backend-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_tracking-backend-native-sam3-1/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=flash_attention_3: inference 11.3 fps, 1.25x the transformers baseline. 2 ids. |
| 2026-10-03 | [tracking-backend-native-sam3.1-sdpa](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-03_tracking-backend-native-sam3-1-sdpa/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3.1 / attn=sdpa: inference 11.4 fps, 1.26x the transformers baseline. 2 ids. |
| 2026-10-02 | [tracking-backend-native-sam3.1](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-backend-native-sam3.1/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | failed | — |
| 2026-10-02 | [tracking-backend-native-sam3.1-sdpa](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-backend-native-sam3.1-sdpa/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | failed | — |
| 2026-10-02 | [tracking-backend-backend-native-compile](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-backend-backend-native-compile/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3 / attn=sdpa / compiled: inference 12.4 fps, 1.37x the transformers baseline. 2 ids. |
| 2026-10-02 | [tracking-backend-backend-native](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-backend-backend-native/summary.md) | How fast is SAM 3 tracking with this backend configuration, compared with the transformers baseline on the same GPU? | ok | native / facebook/sam3 / attn=sdpa: inference 8.3 fps, 0.91x the transformers baseline. 2 ids. |
| 2026-10-02 | [tracking-vs-frames](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-vs-frames-5/summary.md) | On a full clip, how much does SAM 3's tracker cost compared with segmenting every frame independently, and is frame-by-frame segmentation stable enough without it? | ok | Inference only: per-frame 14.6 fps vs tracker 9.0 fps (1.6x); tracker setup 3.6 s, rendering ~9.8 s. Per-frame + IoU matching produced 17 ids (tracker 2) and the count left its usual value on 12/200 frames. Compare the two videos in runs/. |
| 2026-10-02 | [tracking-vs-frames](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-vs-frames-4/summary.md) | On a full clip, how much does SAM 3's tracker cost compared with segmenting every frame independently, and is frame-by-frame segmentation stable enough without it? | ok | Inference only: per-frame 14.7 fps vs tracker 9.0 fps (1.6x); tracker setup 3.6 s, rendering ~9.9 s. Per-frame + IoU matching produced 17 ids (tracker 2) and the count left its usual value on 12/200 frames. Compare the two videos in runs/. |
| 2026-10-02 | [tracking-vs-frames](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-vs-frames-3/summary.md) | On a full clip, how much does SAM 3's tracker cost compared with segmenting every frame independently, and is frame-by-frame segmentation stable enough without it? | ok | Inference only: per-frame 14.6 fps vs tracker 9.1 fps (1.6x); tracker setup 3.6 s, rendering ~9.9 s. Per-frame + IoU matching produced 17 ids (tracker 2) and the count left its usual value on 12/200 frames. Compare the two videos in runs/. |
| 2026-10-02 | [tracking-vs-frames](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-vs-frames-2/summary.md) | On a full clip, how much does SAM 3's tracker cost compared with segmenting every frame independently, and is frame-by-frame segmentation stable enough without it? | ok | Inference only: per-frame 14.4 fps vs tracker 8.8 fps (1.6x); tracker setup 3.5 s, rendering ~9.8 s. Per-frame + IoU matching produced 17 ids (tracker 2) and the count left its usual value on 12/200 frames. Compare the two videos in runs/. |
| 2026-10-02 | [tracking-vs-frames](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_tracking-vs-frames/summary.md) | On a full clip, how much does SAM 3's tracker cost compared with segmenting every frame independently, and what does tracking buy? | ok | Frame-by-frame segmentation ran at 16.1 fps vs the tracker's 5.0 fps (3.2x), i.e. 1.87x vs 6.05x the clip's length. Without tracking, the object count changed on 12 of 200 frames. |
| 2026-10-02 | [backbone-head-split](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-10-02_backbone-head-split/summary.md) | How is per-frame cost split between the vision backbone and the per-prompt head? | ok | The backbone is 49% of per-frame cost. Computing it once per frame and reusing it for every prompt makes each extra object cost ~0.04 s instead of ~0.07 s (1.7x faster for 8 prompts). |
| 2026-09-29 | [backbone-head-split](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-29_backbone-head-split/summary.md) | How is per-frame cost split between the vision backbone and the per-prompt head? | ok | The backbone is 48% of per-frame cost. Computing it once per frame and reusing it for every prompt makes each extra object cost ~0.04 s instead of ~0.07 s (1.7x faster for 8 prompts). |
| 2026-09-24 | [resolution-ladder](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-24_resolution-ladder/summary.md) | Does input resolution change SAM 3's tracking speed, or its ability to find the object? | ok | Speed ranged 0.55-0.61 fps across a 256x pixel range. 'bed' was found in every frame at every size, down to 60x34. |
| 2026-09-24 | [batch-size-sweep](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-24_batch-size-sweep/summary.md) | Does batching frames through the image model (per-frame segmentation, no tracking) raise throughput? | ok | Best was batch 2 at 0.95 fps, 1.06x batch 1. Batching gives no real gain here: one frame already saturates the GPU. |
| 2026-09-24 | [backbone-head-split](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-24_backbone-head-split/summary.md) | How is per-frame cost split between the vision backbone and the per-prompt head? | ok | The backbone is 86% of per-frame cost. Computing it once per frame and reusing it for every prompt makes each extra object cost ~0.11 s instead of ~1.04 s (4.5x faster for 8 prompts). |
| 2026-09-21 | [hardware-m3-vs-t4](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-21_hardware-m3-vs-t4/summary.md) | Is a Colab T4 faster than the M3 Max for SAM 3 video tracking? | reconstructed | No: the T4 is ~1.8x slower than the laptop, and fp16 barely helped -- the T4 is limited by memory bandwidth. L4/A100 are the first GPUs where bf16 and FlashAttention-2 engage. |
| 2026-09-15 | [harness-regression-tests](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-15_harness-regression-tests/summary.md) | After each refactor, do image and video runs still give the same detections, with correct timings? | reconstructed | Detections stayed identical across every refactor. Per-frame timing was wrong until it moved to forward hooks, because the tracker buffers its first 15 frames. |
| 2026-09-15 | [resolution-ladder](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-15_resolution-ladder/summary.md) | Does input resolution change SAM 3's speed, or its ability to find the object? | reconstructed | No: speed is flat at 0.60-0.63 fps because SAM 3 resamples every input to 1008x1008, and the bed was still tracked 40/40 frames at 60x34. |
| 2026-09-12 | [legend-autofit](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-12_legend-autofit/summary.md) | Does the diagnostic legend stay readable and inside the image as image size shrinks? | reconstructed | Originally no -- at 73x110 the panel covered the image. Fixed 9/13 by shrinking the font to fit, then dropping rows. |
| 2026-09-12 | [first-image-tests](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-12_first-image-tests/summary.md) | Does SAM 3 (transformers port, M3 Max) find every instance of a text prompt in a still image? | reconstructed | Yes: all 4 truck wheels were found, including the 2 partly hidden ones, at ~1.1 s per image. |
| 2026-09-12 | [first-tracking-test](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/2026-09-12_first-tracking-test/summary.md) | Can SAM 3 video PCS detect and track a text prompt through a full clip, and how fast on the M3 Max? | reconstructed | Yes, identity held for all 200 frames, but at 0.52 fps -- about 58x slower than real time. |

Runs outside any experiment: [experiments/adhoc](https://huggingface.co/buckets/zachnorton03/llmstreaming/tree/experiments/adhoc)