diogoantunes commited on
Commit
e4b2f9f
·
verified ·
1 Parent(s): f890f05

Release verified Wave 5 checkpoint (seed 2001)

Browse files

Original BF16 and portable INT8 exports with recorded configuration, metrics, checksums, and a validated NumPy loader.

README.md ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ library_name: jax
5
+ tags:
6
+ - world-model
7
+ - multimodal
8
+ - working-memory
9
+ - embodied-control
10
+ - research
11
+ - npz
12
+ ---
13
+
14
+ # SLWM-X1B — Wave 5
15
+
16
+ The 1.008B-parameter signal-latent world-model flagship with doubled training exposure, action-conditioned latent prediction, editable matrix memory, and explicit commitment routing.
17
+
18
+ Research by **Diogo Antunes**, at [valuealigned.ai](https://valuealigned.ai) and [keplen.ai](https://keplen.ai). Published in the **valueai-org** organization.
19
+
20
+ This is an original, unmodified checkpoint from the SLWM research archive. It uses a **custom NumPy NPZ format** and the SLWM JAX runtime; it is not a Transformers `AutoModel` checkpoint or a chat model.
21
+
22
+ ## Exact release
23
+
24
+ | Property | Value |
25
+ |---|---|
26
+ | Original run | `x1b-w5-slwm-1b-s2001` |
27
+ | Experimental wave | 5 |
28
+ | Training seed | 2001 |
29
+ | Strict total parameters | 1,008,454,410 |
30
+ | Training steps | 5,600 |
31
+ | Batch size | 12 sequences |
32
+ | Training sequence length | 32 ticks |
33
+ | Latent width / layers / slots per tick | 1536 / 23 / 24 |
34
+ | Runtime/source reference | [`cbb2b38`](https://github.com/ValueAlignedAI/slwm/tree/cbb2b38794359adfefe07b19e9b7caa113b25b47) |
35
+
36
+ Seed 2001 is released as a fixed reference, not selected by ranking test scores. The research paper reports three-seed cohorts; this repository contains one checkpoint and its own recorded metrics.
37
+
38
+ Wave 5 uses the original observation and central-barrier planning setting, including object coordinates in proprioception. Its physical xArm6 evaluation is offline and trajectory-disjoint within one workspace.
39
+
40
+ ## Files and integrity
41
+
42
+ | File | Bytes | SHA-256 |
43
+ |---|---:|---|
44
+ | `weights_bf16.npz` | 1,602,108,632 | `f56b803857a09e7e77e1bb501efdc8e47bcacb2230cf18a1ab44e1acca591f51` |
45
+ | `portable_int8.npz` | 918,039,628 | `faaf635499f1ba8f501ecd64911ffa0bf88969f6637d521337d45ee4bf1ea36f` |
46
+
47
+ - `weights_bf16.npz`: the project's mixed BF16/FP32 export. Matrix BF16 values are stored as uint16 bit patterns, with a dtype manifest.
48
+ - `portable_int8.npz`: the project's symmetric per-channel INT8 export, with scales and a quantization manifest.
49
+ - `config.json`, `receipt.json`, `eval_metrics.json`, and `train.jsonl`: original run artifacts, preserved byte for byte.
50
+ - `release_manifest.json` and `SHA256SUMS`: source mapping, artifact hashes, and export-validation results.
51
+ - `load_checkpoint.py`: a small, pickle-free NumPy loader for both formats.
52
+
53
+ All original files were checked against their recorded SHA-256 values before upload. Both weight exports were decoded on CPU, checked for finite values, and counted against the strict total and component counts. These are packaging/integrity checks, not new benchmark evaluations.
54
+
55
+ The provided loader expands weights to float32. **INT8 file size is not the runtime memory requirement, and this loader does not perform INT8 matrix computation.** Model weights, activations, and planning buffers are separate from the paper's bounded streaming-state figures.
56
+
57
+ ## Download and load
58
+
59
+ Install `numpy` and `huggingface_hub` in your Python environment, then:
60
+
61
+ ```python
62
+ import json
63
+ import sys
64
+ from pathlib import Path
65
+ from huggingface_hub import snapshot_download
66
+
67
+ folder = Path(snapshot_download(
68
+ repo_id="valueai-org/SLWM-X1B",
69
+ allow_patterns=["portable_int8.npz", "config.json", "receipt.json", "load_checkpoint.py"],
70
+ ))
71
+ sys.path.insert(0, str(folder))
72
+ from load_checkpoint import load_checkpoint
73
+
74
+ receipt = json.loads((folder / "receipt.json").read_text())
75
+ params = load_checkpoint(
76
+ folder / "portable_int8.npz",
77
+ expected_sha256=receipt["portable_int8_sha256"],
78
+ )
79
+ assert sum(value.size for value in params.values()) == receipt["parameter_counts"]["strict_total"]
80
+ ```
81
+
82
+ For the BF16 export, download `weights_bf16.npz` instead and use `receipt["weights_bf16_sha256"]` for verification. `params` is a dictionary of NumPy float32 arrays. JAX execution requires JAX and the appropriate modules from the [pinned research source](https://github.com/ValueAlignedAI/slwm/tree/cbb2b38794359adfefe07b19e9b7caa113b25b47); follow that source for input preparation, modality masks, state updates, and planning.
83
+
84
+ Use `models/x1b_world.py` for the core, `models/x1b_jepa.py` for latent prediction and readouts, and the matching planner/evaluation modules in the pinned source. `scripts/eval/x1b_infer.py` can load local NPZ checkpoints through its `load_verified_checkpoint` function; supply this release directory so the Wave 5 receipt and configuration are used.
85
+
86
+ ## Recorded results for this checkpoint
87
+
88
+ These values come directly from this run's `receipt.json`. They are **single-seed results**, not the manuscript's cohort mean ± standard deviation. The upload does not claim that the quantized export has received an additional independent evaluation.
89
+
90
+ | Recorded metric | Seed 2001 |
91
+ |---|---:|
92
+ | Overall synthetic MPC success | 75.0000 % |
93
+ | One-pixel barrier MPC success | 93.7500 % |
94
+ | T = 76 goal-switch success | 100.0000 % |
95
+ | T = 1,024 associative recall | 100.0000 % |
96
+ | Single-workspace xArm6 four-step error | 7.3368 mm |
97
+
98
+ The original `eval_metrics.json` provides the wider evaluation record, including failures. Compare architectures within the same wave and observation/training setting.
99
+
100
+ ## Scope and limitations
101
+
102
+ This is a research world-model checkpoint, not a general-purpose language or robotics foundation model. Wave 5 control results include privileged object-state inputs and a central-barrier planning prior. The physical-robot measurements are offline trajectory predictions, not live robot deployment. The later scene-disjoint and unprivileged spatial tests belong to different models and waves.
103
+
104
+ The checkpoint is intended for research and reproducibility. The upload contains no raw training recordings, datasets, optimizer state, or private server access configuration. The pinned source commit is a public runtime/release reference, not a separately verified training-code commit: the archived receipts do not record the training Git SHA.
105
+
106
+ ## Sources and licensing
107
+
108
+ - [Original manuscript v0.3 (PDF)](https://diogoantunes.ai/media/publications/slwm_x1b_paper_v03.pdf)
109
+ - [Expanded article, figures, and experiment context](https://diogoantunes.ai/publications/signal-latent-world-models)
110
+ - [SLWM source repository](https://github.com/ValueAlignedAI/slwm/tree/cbb2b38794359adfefe07b19e9b7caa113b25b47)
111
+ - [Original experiment records](https://github.com/ValueAlignedAI/slwm/tree/cbb2b38794359adfefe07b19e9b7caa113b25b47/experiments/multimodal/x1b/spatiotemporal-jepa-1b-20261111/runs_wave5/x1b-w5-slwm-1b-s2001)
112
+
113
+ The source code is distributed under the MIT license reproduced in `SOURCE_LICENSE`. No separate model-weight license was specified in the archived release metadata; this upload does not invent an additional license grant.
114
+
115
+ ## Citation
116
+
117
+ Diogo Antunes. *Signal-Latent World Models at 1.008B Scale: Separating Fast Sensory Grounding from Editable Working Memory Under O(1) Streaming Control.* Preprint, manuscript v0.3, November 14, 2026.
SHA256SUMS ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ c2e84932365c3808e8d0c1b3d6208ebe200b5107fbbbda2df37e04ddcc9b63c7 README.md
2
+ c56a80b248a6b638e49a4365fff816160532b95bb5264a47f9dc67ae8fe64a40 SOURCE_LICENSE
3
+ 2e31e53c6431657da058c6293d3c7b948dfb5b221fe5c4fde1c08ea8c5bd2c00 config.json
4
+ ef8878a15da85fd1e14724e6ad50b6d0a6e6c05f1e1aa673c910f44ff576771b eval_metrics.json
5
+ 27686dfc84db3c70b74999e2ee665df6833b99e380df93d3573c857c9c240ca8 load_checkpoint.py
6
+ faaf635499f1ba8f501ecd64911ffa0bf88969f6637d521337d45ee4bf1ea36f portable_int8.npz
7
+ f495c6897d36ececd3f85fad64b5aa7283878e8002a4652bdf0831d584d21347 receipt.json
8
+ 60c0a8385a1128049d802e46cd4fca71cd0461b13e365c5a828ec95aa5307790 release_manifest.json
9
+ c06d962ea090965c6958b75beffae8b1a8c52ae1a45a35d5b5ab04da0de90c9b requirements.txt
10
+ acb8668b83c886fa6fda8797721ee746cd8552c7def3ab940f8625118e6c084b train.jsonl
11
+ f56b803857a09e7e77e1bb501efdc8e47bcacb2230cf18a1ab44e1acca591f51 weights_bf16.npz
SOURCE_LICENSE ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Diogo Antunes <contact@diogoantunes.ai>
4
+ Copyright (c) 2026 ValueAI <contact@valuealigned.ai>
5
+ Copyright (c) 2026 Vaisys <contact@vaisys.com>
6
+ Copyright (c) 2026 Splendor <contact@splendor-os.org>
7
+ Copyright (c) 2026 UniversoAI <contact@universo-ai.com>
8
+
9
+ Permission is hereby granted, free of charge, to any person obtaining a copy
10
+ of this software and associated documentation files (the "Software"), to deal
11
+ in the Software without restriction, including without limitation the rights
12
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
13
+ copies of the Software, and to permit persons to whom the Software is
14
+ furnished to do so, subject to the following conditions:
15
+
16
+ The above copyright notice and this permission notice shall be included in all
17
+ copies or substantial portions of the Software.
18
+
19
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
20
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
22
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
23
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
24
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
25
+ SOFTWARE.
config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tpu",
3
+ "batch_size": 12,
4
+ "config_sha256": "cca9e2ca4e9ccde614b3261a59d939953218168d70af3bae213952652c02087a",
5
+ "devices": [
6
+ "TPU_0(process=0,(0,0,0,0))"
7
+ ],
8
+ "objective": "residual_ema_jepa_sigreg",
9
+ "parameter_counts": {
10
+ "adapters": 19033728,
11
+ "jepa_and_uncertainty": 16529666,
12
+ "policy_and_heads": 18272296,
13
+ "processor_core": 954618720,
14
+ "strict_total": 1008454410
15
+ },
16
+ "peak_lr": 0.00024,
17
+ "real_bank_path": "/content/x1b-data/x1b_real_bank_v1.npz",
18
+ "real_robot_path": "/content/x1b-data/x1b_real_robot_v2.npz",
19
+ "seed": 2001,
20
+ "seq_len": 32,
21
+ "spec": {
22
+ "action_dim": 4,
23
+ "arch": "slwm_x1b",
24
+ "audio_dim": 64,
25
+ "d_ff": 4096,
26
+ "d_model": 1536,
27
+ "dtype": "bfloat16",
28
+ "goal_dim": 4,
29
+ "grad_checkpoint": true,
30
+ "head_dim": 64,
31
+ "layers": 23,
32
+ "n_heads": 24,
33
+ "profile": "x1b",
34
+ "proprio_dim": 16,
35
+ "rope_Window": 512,
36
+ "slots_per_tick": 24,
37
+ "vocab_size": 4096
38
+ },
39
+ "steps": 5600,
40
+ "tag": "x1b-w5-slwm-1b-s2001"
41
+ }
eval_metrics.json ADDED
@@ -0,0 +1,1659 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "associative_recall_512": {
3
+ "T1024": {
4
+ "ci95_high": 1.0,
5
+ "ci95_low": 1.0,
6
+ "mean": 1.0
7
+ },
8
+ "T1024_vec": [
9
+ 1.0,
10
+ 1.0,
11
+ 1.0,
12
+ 1.0,
13
+ 1.0,
14
+ 1.0,
15
+ 1.0,
16
+ 1.0,
17
+ 1.0,
18
+ 1.0,
19
+ 1.0,
20
+ 1.0,
21
+ 1.0,
22
+ 1.0,
23
+ 1.0,
24
+ 1.0
25
+ ],
26
+ "T128": {
27
+ "ci95_high": 1.0,
28
+ "ci95_low": 1.0,
29
+ "mean": 1.0
30
+ },
31
+ "T128_vec": [
32
+ 1.0,
33
+ 1.0,
34
+ 1.0,
35
+ 1.0,
36
+ 1.0,
37
+ 1.0,
38
+ 1.0,
39
+ 1.0,
40
+ 1.0,
41
+ 1.0,
42
+ 1.0,
43
+ 1.0,
44
+ 1.0,
45
+ 1.0,
46
+ 1.0,
47
+ 1.0,
48
+ 1.0,
49
+ 1.0,
50
+ 1.0,
51
+ 1.0,
52
+ 1.0,
53
+ 1.0,
54
+ 1.0,
55
+ 1.0,
56
+ 1.0,
57
+ 1.0,
58
+ 1.0,
59
+ 1.0,
60
+ 1.0,
61
+ 1.0,
62
+ 1.0,
63
+ 1.0
64
+ ],
65
+ "T256": {
66
+ "ci95_high": 1.0,
67
+ "ci95_low": 1.0,
68
+ "mean": 1.0
69
+ },
70
+ "T256_vec": [
71
+ 1.0,
72
+ 1.0,
73
+ 1.0,
74
+ 1.0,
75
+ 1.0,
76
+ 1.0,
77
+ 1.0,
78
+ 1.0,
79
+ 1.0,
80
+ 1.0,
81
+ 1.0,
82
+ 1.0,
83
+ 1.0,
84
+ 1.0,
85
+ 1.0,
86
+ 1.0
87
+ ],
88
+ "T32": {
89
+ "ci95_high": 1.0,
90
+ "ci95_low": 1.0,
91
+ "mean": 1.0
92
+ },
93
+ "T32_vec": [
94
+ 1.0,
95
+ 1.0,
96
+ 1.0,
97
+ 1.0,
98
+ 1.0,
99
+ 1.0,
100
+ 1.0,
101
+ 1.0,
102
+ 1.0,
103
+ 1.0,
104
+ 1.0,
105
+ 1.0,
106
+ 1.0,
107
+ 1.0,
108
+ 1.0,
109
+ 1.0,
110
+ 1.0,
111
+ 1.0,
112
+ 1.0,
113
+ 1.0,
114
+ 1.0,
115
+ 1.0,
116
+ 1.0,
117
+ 1.0,
118
+ 1.0,
119
+ 1.0,
120
+ 1.0,
121
+ 1.0,
122
+ 1.0,
123
+ 1.0,
124
+ 1.0,
125
+ 1.0
126
+ ],
127
+ "T512": {
128
+ "ci95_high": 1.0,
129
+ "ci95_low": 1.0,
130
+ "mean": 1.0
131
+ },
132
+ "T512_vec": [
133
+ 1.0,
134
+ 1.0,
135
+ 1.0,
136
+ 1.0,
137
+ 1.0,
138
+ 1.0,
139
+ 1.0,
140
+ 1.0,
141
+ 1.0,
142
+ 1.0,
143
+ 1.0,
144
+ 1.0,
145
+ 1.0,
146
+ 1.0,
147
+ 1.0,
148
+ 1.0
149
+ ]
150
+ },
151
+ "closed_loop_mpc": {
152
+ "g8_random_shooting_3": {
153
+ "final_distance": {
154
+ "ci95_high": 0.4635117736596537,
155
+ "ci95_low": 0.3840821622987278,
156
+ "mean": 0.4226892536583667
157
+ },
158
+ "pusht_open_success": 0.4791666567325592,
159
+ "succ_vec": [
160
+ 0.0,
161
+ 1.0,
162
+ 0.0,
163
+ 0.0,
164
+ 0.0,
165
+ 0.0,
166
+ 0.0,
167
+ 1.0,
168
+ 0.0,
169
+ 0.0,
170
+ 0.0,
171
+ 0.0,
172
+ 0.0,
173
+ 1.0,
174
+ 0.0,
175
+ 0.0,
176
+ 0.0,
177
+ 0.0,
178
+ 0.0,
179
+ 1.0,
180
+ 0.0,
181
+ 0.0,
182
+ 0.0,
183
+ 1.0,
184
+ 0.0,
185
+ 0.0,
186
+ 0.0,
187
+ 0.0,
188
+ 0.0,
189
+ 1.0,
190
+ 0.0,
191
+ 1.0,
192
+ 0.0,
193
+ 0.0,
194
+ 0.0,
195
+ 0.0,
196
+ 0.0,
197
+ 0.0,
198
+ 0.0,
199
+ 1.0,
200
+ 0.0,
201
+ 0.0,
202
+ 0.0,
203
+ 1.0,
204
+ 0.0,
205
+ 1.0,
206
+ 0.0,
207
+ 1.0,
208
+ 0.0,
209
+ 1.0,
210
+ 0.0,
211
+ 0.0,
212
+ 0.0,
213
+ 1.0,
214
+ 0.0,
215
+ 0.0,
216
+ 0.0,
217
+ 1.0,
218
+ 0.0,
219
+ 0.0,
220
+ 0.0,
221
+ 1.0,
222
+ 0.0,
223
+ 1.0,
224
+ 0.0,
225
+ 0.0,
226
+ 0.0,
227
+ 1.0,
228
+ 0.0,
229
+ 0.0,
230
+ 0.0,
231
+ 1.0,
232
+ 0.0,
233
+ 1.0,
234
+ 0.0,
235
+ 0.0,
236
+ 0.0,
237
+ 1.0,
238
+ 0.0,
239
+ 0.0,
240
+ 0.0,
241
+ 0.0,
242
+ 0.0,
243
+ 1.0,
244
+ 0.0,
245
+ 0.0,
246
+ 0.0,
247
+ 0.0,
248
+ 0.0,
249
+ 0.0,
250
+ 0.0,
251
+ 1.0,
252
+ 0.0,
253
+ 0.0,
254
+ 0.0,
255
+ 1.0
256
+ ],
257
+ "success_rate": {
258
+ "ci95_high": 0.3229166666666667,
259
+ "ci95_low": 0.15625,
260
+ "mean": 0.23958333333333334
261
+ },
262
+ "thin_barrier_success": 0.0
263
+ },
264
+ "latent_cem_mpc": {
265
+ "final_distance": {
266
+ "ci95_high": 0.16836298348377265,
267
+ "ci95_low": 0.1147795234862618,
268
+ "mean": 0.14105345892797536
269
+ },
270
+ "pusht_open_success": 0.5625,
271
+ "succ_vec": [
272
+ 1.0,
273
+ 1.0,
274
+ 1.0,
275
+ 1.0,
276
+ 1.0,
277
+ 0.0,
278
+ 1.0,
279
+ 0.0,
280
+ 1.0,
281
+ 1.0,
282
+ 1.0,
283
+ 0.0,
284
+ 1.0,
285
+ 1.0,
286
+ 1.0,
287
+ 1.0,
288
+ 1.0,
289
+ 1.0,
290
+ 1.0,
291
+ 1.0,
292
+ 0.0,
293
+ 1.0,
294
+ 1.0,
295
+ 0.0,
296
+ 1.0,
297
+ 1.0,
298
+ 1.0,
299
+ 0.0,
300
+ 1.0,
301
+ 1.0,
302
+ 1.0,
303
+ 0.0,
304
+ 1.0,
305
+ 1.0,
306
+ 1.0,
307
+ 0.0,
308
+ 1.0,
309
+ 1.0,
310
+ 1.0,
311
+ 1.0,
312
+ 1.0,
313
+ 1.0,
314
+ 1.0,
315
+ 1.0,
316
+ 1.0,
317
+ 0.0,
318
+ 1.0,
319
+ 1.0,
320
+ 1.0,
321
+ 0.0,
322
+ 1.0,
323
+ 0.0,
324
+ 1.0,
325
+ 1.0,
326
+ 1.0,
327
+ 1.0,
328
+ 0.0,
329
+ 0.0,
330
+ 1.0,
331
+ 0.0,
332
+ 1.0,
333
+ 1.0,
334
+ 1.0,
335
+ 1.0,
336
+ 1.0,
337
+ 1.0,
338
+ 1.0,
339
+ 1.0,
340
+ 1.0,
341
+ 0.0,
342
+ 0.0,
343
+ 1.0,
344
+ 1.0,
345
+ 0.0,
346
+ 1.0,
347
+ 1.0,
348
+ 1.0,
349
+ 1.0,
350
+ 1.0,
351
+ 0.0,
352
+ 1.0,
353
+ 1.0,
354
+ 1.0,
355
+ 0.0,
356
+ 1.0,
357
+ 0.0,
358
+ 1.0,
359
+ 0.0,
360
+ 1.0,
361
+ 0.0,
362
+ 1.0,
363
+ 0.0,
364
+ 1.0,
365
+ 1.0,
366
+ 1.0,
367
+ 0.0
368
+ ],
369
+ "success_rate": {
370
+ "ci95_high": 0.8333333333333334,
371
+ "ci95_low": 0.6666666666666666,
372
+ "mean": 0.75
373
+ },
374
+ "thin_barrier_success": 0.9375
375
+ },
376
+ "open_loop_policy": {
377
+ "final_distance": {
378
+ "ci95_high": 0.40534411281502497,
379
+ "ci95_low": 0.3069167479397341,
380
+ "mean": 0.3551293204460914
381
+ },
382
+ "pusht_open_success": 0.3958333432674408,
383
+ "succ_vec": [
384
+ 0.0,
385
+ 1.0,
386
+ 1.0,
387
+ 1.0,
388
+ 1.0,
389
+ 0.0,
390
+ 1.0,
391
+ 0.0,
392
+ 0.0,
393
+ 1.0,
394
+ 0.0,
395
+ 0.0,
396
+ 0.0,
397
+ 0.0,
398
+ 0.0,
399
+ 0.0,
400
+ 0.0,
401
+ 1.0,
402
+ 0.0,
403
+ 1.0,
404
+ 1.0,
405
+ 1.0,
406
+ 1.0,
407
+ 0.0,
408
+ 1.0,
409
+ 0.0,
410
+ 1.0,
411
+ 0.0,
412
+ 1.0,
413
+ 1.0,
414
+ 1.0,
415
+ 1.0,
416
+ 0.0,
417
+ 0.0,
418
+ 0.0,
419
+ 0.0,
420
+ 0.0,
421
+ 1.0,
422
+ 1.0,
423
+ 1.0,
424
+ 0.0,
425
+ 1.0,
426
+ 1.0,
427
+ 1.0,
428
+ 0.0,
429
+ 0.0,
430
+ 0.0,
431
+ 1.0,
432
+ 1.0,
433
+ 0.0,
434
+ 1.0,
435
+ 0.0,
436
+ 1.0,
437
+ 0.0,
438
+ 1.0,
439
+ 0.0,
440
+ 1.0,
441
+ 0.0,
442
+ 1.0,
443
+ 1.0,
444
+ 1.0,
445
+ 1.0,
446
+ 0.0,
447
+ 0.0,
448
+ 1.0,
449
+ 0.0,
450
+ 0.0,
451
+ 1.0,
452
+ 0.0,
453
+ 0.0,
454
+ 1.0,
455
+ 1.0,
456
+ 1.0,
457
+ 0.0,
458
+ 1.0,
459
+ 1.0,
460
+ 0.0,
461
+ 0.0,
462
+ 1.0,
463
+ 0.0,
464
+ 0.0,
465
+ 1.0,
466
+ 1.0,
467
+ 0.0,
468
+ 1.0,
469
+ 0.0,
470
+ 1.0,
471
+ 0.0,
472
+ 1.0,
473
+ 0.0,
474
+ 1.0,
475
+ 0.0,
476
+ 1.0,
477
+ 0.0,
478
+ 0.0,
479
+ 0.0
480
+ ],
481
+ "success_rate": {
482
+ "ci95_high": 0.59375,
483
+ "ci95_low": 0.40625,
484
+ "mean": 0.5
485
+ },
486
+ "thin_barrier_success": 0.6041666865348816
487
+ }
488
+ },
489
+ "long_horizon_memory_mpc": {
490
+ "mid_episode_goal_switch_mpc": {
491
+ "final_distance": {
492
+ "ci95_high": 0.10044096388010076,
493
+ "ci95_low": 0.07583995317254448,
494
+ "mean": 0.0879217807087116
495
+ },
496
+ "open_success": 1.0,
497
+ "stale_goal_a_trap_rate": 0.0,
498
+ "succ_vec": [
499
+ 1.0,
500
+ 1.0,
501
+ 1.0,
502
+ 1.0,
503
+ 1.0,
504
+ 1.0,
505
+ 1.0,
506
+ 1.0,
507
+ 1.0,
508
+ 1.0,
509
+ 1.0,
510
+ 1.0,
511
+ 1.0,
512
+ 1.0,
513
+ 1.0,
514
+ 1.0,
515
+ 1.0,
516
+ 1.0,
517
+ 1.0,
518
+ 1.0,
519
+ 1.0,
520
+ 1.0,
521
+ 1.0,
522
+ 1.0,
523
+ 1.0,
524
+ 1.0,
525
+ 1.0,
526
+ 1.0,
527
+ 1.0,
528
+ 1.0,
529
+ 1.0,
530
+ 1.0,
531
+ 1.0,
532
+ 1.0,
533
+ 1.0,
534
+ 1.0,
535
+ 1.0,
536
+ 1.0,
537
+ 1.0,
538
+ 1.0,
539
+ 1.0,
540
+ 1.0,
541
+ 1.0,
542
+ 1.0,
543
+ 1.0,
544
+ 1.0,
545
+ 1.0,
546
+ 1.0,
547
+ 1.0,
548
+ 1.0,
549
+ 1.0,
550
+ 1.0,
551
+ 1.0,
552
+ 1.0,
553
+ 1.0,
554
+ 1.0,
555
+ 1.0,
556
+ 1.0,
557
+ 1.0,
558
+ 1.0,
559
+ 1.0,
560
+ 1.0,
561
+ 1.0,
562
+ 1.0
563
+ ],
564
+ "success_rate": {
565
+ "ci95_high": 1.0,
566
+ "ci95_low": 1.0,
567
+ "mean": 1.0
568
+ },
569
+ "thin_barrier_success": 1.0
570
+ },
571
+ "sustained_transient_barrier_mpc": {
572
+ "final_distance": {
573
+ "ci95_high": 0.05832585472162464,
574
+ "ci95_low": 0.04480057375258184,
575
+ "mean": 0.05143831421446521
576
+ },
577
+ "open_success": 1.0,
578
+ "succ_vec": [
579
+ 1.0,
580
+ 1.0,
581
+ 1.0,
582
+ 1.0,
583
+ 1.0,
584
+ 1.0,
585
+ 1.0,
586
+ 1.0,
587
+ 1.0,
588
+ 1.0,
589
+ 1.0,
590
+ 1.0,
591
+ 1.0,
592
+ 1.0,
593
+ 1.0,
594
+ 1.0,
595
+ 1.0,
596
+ 1.0,
597
+ 1.0,
598
+ 1.0,
599
+ 1.0,
600
+ 1.0,
601
+ 1.0,
602
+ 1.0,
603
+ 1.0,
604
+ 1.0,
605
+ 1.0,
606
+ 1.0,
607
+ 1.0,
608
+ 1.0,
609
+ 1.0,
610
+ 1.0,
611
+ 1.0,
612
+ 1.0,
613
+ 1.0,
614
+ 1.0,
615
+ 1.0,
616
+ 1.0,
617
+ 1.0,
618
+ 1.0,
619
+ 1.0,
620
+ 1.0,
621
+ 1.0,
622
+ 1.0,
623
+ 1.0,
624
+ 1.0,
625
+ 1.0,
626
+ 1.0,
627
+ 1.0,
628
+ 1.0,
629
+ 1.0,
630
+ 1.0,
631
+ 1.0,
632
+ 1.0,
633
+ 1.0,
634
+ 1.0,
635
+ 1.0,
636
+ 1.0,
637
+ 1.0,
638
+ 1.0,
639
+ 1.0,
640
+ 1.0,
641
+ 1.0,
642
+ 1.0
643
+ ],
644
+ "success_rate": {
645
+ "ci95_high": 1.0,
646
+ "ci95_low": 1.0,
647
+ "mean": 1.0
648
+ },
649
+ "thin_barrier_success": 1.0
650
+ }
651
+ },
652
+ "real_multimodal_and_text": {
653
+ "av_r1": {
654
+ "ci95_high": 0.984375,
655
+ "ci95_low": 0.84375,
656
+ "mean": 0.921875
657
+ },
658
+ "av_r5": {
659
+ "ci95_high": 1.0,
660
+ "ci95_low": 1.0,
661
+ "mean": 1.0
662
+ },
663
+ "available": true,
664
+ "contact_auroc": 0.991238775312514,
665
+ "fineweb_final_acc": 0.125,
666
+ "fineweb_nll": 6.74179389542763,
667
+ "fineweb_ppl": 847.0789447357108,
668
+ "participation_ratio": 0.035112855532629925,
669
+ "per_item_r1": [
670
+ 1.0,
671
+ 1.0,
672
+ 1.0,
673
+ 1.0,
674
+ 1.0,
675
+ 1.0,
676
+ 1.0,
677
+ 1.0,
678
+ 1.0,
679
+ 1.0,
680
+ 1.0,
681
+ 1.0,
682
+ 1.0,
683
+ 1.0,
684
+ 1.0,
685
+ 1.0,
686
+ 1.0,
687
+ 1.0,
688
+ 1.0,
689
+ 0.0,
690
+ 1.0,
691
+ 1.0,
692
+ 1.0,
693
+ 0.0,
694
+ 1.0,
695
+ 1.0,
696
+ 1.0,
697
+ 1.0,
698
+ 1.0,
699
+ 1.0,
700
+ 1.0,
701
+ 1.0,
702
+ 1.0,
703
+ 1.0,
704
+ 1.0,
705
+ 1.0,
706
+ 1.0,
707
+ 1.0,
708
+ 1.0,
709
+ 1.0,
710
+ 1.0,
711
+ 0.0,
712
+ 1.0,
713
+ 1.0,
714
+ 1.0,
715
+ 0.0,
716
+ 1.0,
717
+ 1.0,
718
+ 1.0,
719
+ 1.0,
720
+ 1.0,
721
+ 1.0,
722
+ 1.0,
723
+ 1.0,
724
+ 1.0,
725
+ 1.0,
726
+ 1.0,
727
+ 1.0,
728
+ 1.0,
729
+ 1.0,
730
+ 0.0,
731
+ 1.0,
732
+ 1.0,
733
+ 1.0
734
+ ],
735
+ "per_item_sq_err": [
736
+ 0.0030270263087004423,
737
+ 1.4591902072425e-05,
738
+ 8.993362280307338e-05,
739
+ 3.458862556726672e-05,
740
+ 6.619398482143879e-05,
741
+ 7.252775685628876e-05,
742
+ 0.00013622491678688675,
743
+ 0.00015096009883563966,
744
+ 1.589155181136448e-05,
745
+ 0.0030661558266729116,
746
+ 0.0001850870467023924,
747
+ 2.4606912120361812e-05,
748
+ 3.169229967170395e-05,
749
+ 4.968203211319633e-05,
750
+ 0.00036145257763564587,
751
+ 2.257074993394781e-05,
752
+ 3.448061033850536e-05,
753
+ 0.000215750013012439,
754
+ 0.00046554123400710523,
755
+ 0.0006946134380996227,
756
+ 2.259343455079943e-05,
757
+ 0.00013055730960331857,
758
+ 0.0008733897702768445,
759
+ 0.00035739815211854875,
760
+ 1.1370071661076508e-05,
761
+ 2.8524382287287153e-05,
762
+ 5.137251355336048e-05,
763
+ 0.00014816835755482316,
764
+ 0.0003815449890680611,
765
+ 2.5599343643989414e-05,
766
+ 0.00023317625164054334,
767
+ 0.00039288270636461675,
768
+ 6.633723387494683e-05,
769
+ 7.838199962861836e-05,
770
+ 6.908349314471707e-05,
771
+ 0.00013352501264307648,
772
+ 5.4764270316809416e-05,
773
+ 0.0006541479378938675,
774
+ 0.000541939924005419,
775
+ 4.214028376736678e-05,
776
+ 4.264468952897005e-05,
777
+ 4.0836916014086455e-05,
778
+ 0.00038086954737082124,
779
+ 6.647428381256759e-05,
780
+ 3.68979704035155e-06,
781
+ 3.473301330814138e-05,
782
+ 0.0022212681360542774,
783
+ 0.0012026653857901692,
784
+ 0.0014583839802071452,
785
+ 0.00011980589624727145,
786
+ 3.0443232390098274e-05,
787
+ 9.979685273719952e-05,
788
+ 0.00016306359611917287,
789
+ 0.00014964642468839884,
790
+ 4.981197707820684e-05,
791
+ 0.00020226053311489522,
792
+ 0.00015880515275057405,
793
+ 1.8976785213453695e-05,
794
+ 0.00043096786248497665,
795
+ 2.7003319701179862e-05,
796
+ 0.0015155666042119265,
797
+ 0.002033452270552516,
798
+ 8.179085853043944e-05,
799
+ 5.100488124298863e-06,
800
+ 0.0001868645049398765,
801
+ 0.0015103068435564637,
802
+ 4.976815762347542e-05,
803
+ 0.00014845217810943723,
804
+ 0.00010903556540142745,
805
+ 3.1882565963314846e-05,
806
+ 0.000661217316519469,
807
+ 4.271894067642279e-05,
808
+ 0.0007022933568805456,
809
+ 8.116241951938719e-05,
810
+ 5.248594607110135e-05,
811
+ 0.0005650909151881933,
812
+ 0.00024562160251662135,
813
+ 4.088843706995249e-05,
814
+ 8.19546403363347e-06,
815
+ 0.00011863988765981048,
816
+ 0.001141895423643291,
817
+ 0.00029162695864215493,
818
+ 0.00014452530012931675,
819
+ 1.1566550710995216e-05,
820
+ 0.00023995802621357143,
821
+ 0.0013748795026913285,
822
+ 5.278869139146991e-05,
823
+ 0.00033971676020883024,
824
+ 5.1929819164797664e-05,
825
+ 0.0015989179955795407,
826
+ 5.332433647708967e-05,
827
+ 0.0009928473737090826,
828
+ 2.3574270016979426e-05,
829
+ 0.0017843358218669891,
830
+ 3.524993735481985e-05,
831
+ 1.5197337233985309e-05,
832
+ 2.4329112420673482e-05,
833
+ 8.908420568332076e-05,
834
+ 6.127365486463532e-05,
835
+ 0.001333534368313849,
836
+ 0.0008624670444987714,
837
+ 3.545829531503841e-05,
838
+ 4.0993854781845585e-05,
839
+ 0.0005459789535962045,
840
+ 0.0005899496027268469,
841
+ 4.339451334089972e-05,
842
+ 3.873184687108733e-05,
843
+ 0.00017525684961583465,
844
+ 0.0023251415695995092,
845
+ 0.00026077034999616444,
846
+ 0.00022918607282917947,
847
+ 4.643265492632054e-05,
848
+ 2.9291637474671006e-05,
849
+ 0.0013106183614581823,
850
+ 8.670848183101043e-05,
851
+ 0.0007372063701041043,
852
+ 0.0005013621994294226,
853
+ 5.076083107269369e-05,
854
+ 0.0007262479630298913,
855
+ 0.0001326856145169586,
856
+ 1.1932410416193306e-05,
857
+ 2.393837166891899e-05,
858
+ 0.00041041328222490847,
859
+ 0.0007084122044034302,
860
+ 0.0007039483170956373,
861
+ 0.0003529046953190118,
862
+ 0.0026884661056101322,
863
+ 2.9264690965646878e-05,
864
+ 1.54452609422151e-05,
865
+ 0.0002019769890466705,
866
+ 0.0009504136978648603,
867
+ 4.863069989369251e-05,
868
+ 7.642935088369995e-05,
869
+ 0.0014863166725263,
870
+ 0.0002015460340771824,
871
+ 9.462448360864073e-05,
872
+ 0.002174781635403633,
873
+ 2.147544546460267e-05,
874
+ 0.0008840746595524251,
875
+ 0.0002918824611697346,
876
+ 5.993877493892796e-05,
877
+ 6.012023368384689e-05,
878
+ 3.9308328268816695e-05,
879
+ 7.16616923455149e-05,
880
+ 3.8645524909952655e-05,
881
+ 5.033581328461878e-05,
882
+ 0.00023973525094334036,
883
+ 2.6037623683805577e-05,
884
+ 0.0007236427045427263,
885
+ 0.0002748020051512867,
886
+ 7.227355672512203e-05,
887
+ 7.585775892948732e-05,
888
+ 0.00019503181101754308,
889
+ 1.268740015802905e-05,
890
+ 3.917247522622347e-05,
891
+ 5.548095214180648e-05,
892
+ 0.00010281213326379657,
893
+ 0.0009287442080676556,
894
+ 0.0007020283374004066,
895
+ 6.251616287045181e-05,
896
+ 0.0001225015294039622,
897
+ 3.081419345107861e-05,
898
+ 3.0581966711906716e-05,
899
+ 0.00012533151311799884,
900
+ 6.914323876117123e-06,
901
+ 2.529459015931934e-05,
902
+ 8.14376980997622e-05,
903
+ 7.13579574949108e-05,
904
+ 3.543309139786288e-05,
905
+ 0.0007693312363699079,
906
+ 6.624152592848986e-05,
907
+ 0.001110514742322266,
908
+ 0.000198714216821827,
909
+ 0.00015030965732876211,
910
+ 2.4887898689485155e-05,
911
+ 0.0001225033338414505,
912
+ 5.120917558087967e-05,
913
+ 3.332079722895287e-05,
914
+ 9.126830991590396e-05,
915
+ 9.134651918429881e-05,
916
+ 0.0011830513831228018,
917
+ 2.4823779313010164e-05,
918
+ 3.784508226090111e-05,
919
+ 5.828586290590465e-05,
920
+ 9.762363333720714e-05,
921
+ 0.0019706867169588804,
922
+ 1.661188252910506e-05,
923
+ 0.000743208802305162,
924
+ 6.904463225509971e-05,
925
+ 0.0006614333251491189,
926
+ 5.2649185818154365e-05,
927
+ 0.0008287769742310047,
928
+ 6.669304639217444e-06,
929
+ 0.0017017011996358633,
930
+ 2.282290006405674e-05,
931
+ 2.573643359937705e-05,
932
+ 2.5963110601878725e-05,
933
+ 0.0005657850997522473,
934
+ 6.997735181357712e-05,
935
+ 6.224519165698439e-05,
936
+ 6.294205377344042e-05,
937
+ 0.00017129168554674834,
938
+ 4.428517786436714e-05,
939
+ 0.0013409196399152279,
940
+ 2.8377411581459455e-05,
941
+ 0.000287050090264529,
942
+ 5.0210204790346324e-05,
943
+ 0.00011467467265902087,
944
+ 0.00012079588486813009,
945
+ 6.995461444603279e-05,
946
+ 0.0010747900232672691,
947
+ 0.0006124479114077985,
948
+ 0.0012840167619287968,
949
+ 4.011585042462684e-05,
950
+ 8.400646038353443e-05,
951
+ 0.003675664309412241,
952
+ 0.0017483438132330775,
953
+ 7.476330210920423e-05,
954
+ 3.172240758431144e-05,
955
+ 0.0007087328121997416,
956
+ 3.109695535385981e-05,
957
+ 3.915245906682685e-05,
958
+ 0.0009928604122251272,
959
+ 4.581959728966467e-05,
960
+ 6.488135841209441e-05,
961
+ 3.864155587507412e-05,
962
+ 4.8650905227987096e-05,
963
+ 4.565911149256863e-05,
964
+ 0.00011904251005034894,
965
+ 0.0018660288769751787,
966
+ 0.00022033898858353496,
967
+ 6.101536928326823e-05,
968
+ 0.00016235957446042448,
969
+ 1.6646859876345843e-05,
970
+ 0.0008472055196762085,
971
+ 0.0002450648753438145,
972
+ 6.950330134714022e-05,
973
+ 5.787532791146077e-05,
974
+ 1.7140320778707974e-05,
975
+ 0.0006838078261353076,
976
+ 2.6330684704589657e-05,
977
+ 0.00015543682093266398,
978
+ 1.7643200408201665e-05,
979
+ 0.0007158586522564292,
980
+ 0.000244716095039621,
981
+ 8.798262570053339e-05,
982
+ 0.00032641555299051106,
983
+ 0.0009544502827338874,
984
+ 0.0003018635034095496,
985
+ 0.00013014614523854107,
986
+ 0.0002702752244658768,
987
+ 0.00010897059109993279,
988
+ 0.000515399151481688,
989
+ 8.022553083719686e-05,
990
+ 0.00010950559953926131,
991
+ 1.029856412060326e-05
992
+ ],
993
+ "persistence_rmse": 0.021005449816584587,
994
+ "persistence_skill": 0.08482918876690748,
995
+ "real_forecast_rmse": 0.019223574548959732
996
+ },
997
+ "real_robot_transfer": {
998
+ "available": true,
999
+ "causal_plan_gain_mm": {
1000
+ "ci95_high": 170.00423828917556,
1001
+ "ci95_low": 155.0073539149249,
1002
+ "mean": 162.56198576372117
1003
+ },
1004
+ "command_discrimination_acc": {
1005
+ "ci95_high": 0.6171875,
1006
+ "ci95_low": 0.49609375,
1007
+ "mean": 0.55859375
1008
+ },
1009
+ "counterfactual_endpoint_gain_mm": {
1010
+ "ci95_high": 24.851032691076398,
1011
+ "ci95_low": 4.667875496111811,
1012
+ "mean": 14.730599962174892
1013
+ },
1014
+ "disc_vec": [
1015
+ 1.0,
1016
+ 0.0,
1017
+ 1.0,
1018
+ 0.0,
1019
+ 0.0,
1020
+ 0.0,
1021
+ 0.0,
1022
+ 1.0,
1023
+ 0.0,
1024
+ 1.0,
1025
+ 1.0,
1026
+ 1.0,
1027
+ 0.0,
1028
+ 1.0,
1029
+ 0.0,
1030
+ 0.0,
1031
+ 0.0,
1032
+ 0.0,
1033
+ 1.0,
1034
+ 0.0,
1035
+ 1.0,
1036
+ 1.0,
1037
+ 0.0,
1038
+ 0.0,
1039
+ 0.0,
1040
+ 1.0,
1041
+ 0.0,
1042
+ 0.0,
1043
+ 1.0,
1044
+ 1.0,
1045
+ 0.0,
1046
+ 0.0,
1047
+ 0.0,
1048
+ 1.0,
1049
+ 1.0,
1050
+ 0.0,
1051
+ 1.0,
1052
+ 0.0,
1053
+ 1.0,
1054
+ 1.0,
1055
+ 0.0,
1056
+ 1.0,
1057
+ 1.0,
1058
+ 0.0,
1059
+ 1.0,
1060
+ 1.0,
1061
+ 0.0,
1062
+ 1.0,
1063
+ 1.0,
1064
+ 1.0,
1065
+ 1.0,
1066
+ 0.0,
1067
+ 0.0,
1068
+ 0.0,
1069
+ 1.0,
1070
+ 0.0,
1071
+ 0.0,
1072
+ 0.0,
1073
+ 1.0,
1074
+ 1.0,
1075
+ 1.0,
1076
+ 0.0,
1077
+ 0.0,
1078
+ 0.0,
1079
+ 0.0,
1080
+ 0.0,
1081
+ 0.0,
1082
+ 1.0,
1083
+ 1.0,
1084
+ 0.0,
1085
+ 0.0,
1086
+ 1.0,
1087
+ 1.0,
1088
+ 0.0,
1089
+ 1.0,
1090
+ 1.0,
1091
+ 1.0,
1092
+ 1.0,
1093
+ 1.0,
1094
+ 0.0,
1095
+ 1.0,
1096
+ 1.0,
1097
+ 0.0,
1098
+ 1.0,
1099
+ 1.0,
1100
+ 1.0,
1101
+ 1.0,
1102
+ 1.0,
1103
+ 1.0,
1104
+ 1.0,
1105
+ 0.0,
1106
+ 0.0,
1107
+ 1.0,
1108
+ 1.0,
1109
+ 0.0,
1110
+ 0.0,
1111
+ 0.0,
1112
+ 0.0,
1113
+ 0.0,
1114
+ 0.0,
1115
+ 1.0,
1116
+ 1.0,
1117
+ 0.0,
1118
+ 0.0,
1119
+ 1.0,
1120
+ 0.0,
1121
+ 1.0,
1122
+ 1.0,
1123
+ 1.0,
1124
+ 1.0,
1125
+ 1.0,
1126
+ 0.0,
1127
+ 0.0,
1128
+ 1.0,
1129
+ 0.0,
1130
+ 1.0,
1131
+ 1.0,
1132
+ 0.0,
1133
+ 0.0,
1134
+ 1.0,
1135
+ 1.0,
1136
+ 0.0,
1137
+ 1.0,
1138
+ 0.0,
1139
+ 1.0,
1140
+ 1.0,
1141
+ 0.0,
1142
+ 0.0,
1143
+ 0.0,
1144
+ 1.0,
1145
+ 1.0,
1146
+ 1.0,
1147
+ 0.0,
1148
+ 1.0,
1149
+ 0.0,
1150
+ 0.0,
1151
+ 0.0,
1152
+ 0.0,
1153
+ 0.0,
1154
+ 1.0,
1155
+ 0.0,
1156
+ 0.0,
1157
+ 1.0,
1158
+ 0.0,
1159
+ 1.0,
1160
+ 1.0,
1161
+ 0.0,
1162
+ 1.0,
1163
+ 0.0,
1164
+ 0.0,
1165
+ 1.0,
1166
+ 1.0,
1167
+ 1.0,
1168
+ 1.0,
1169
+ 0.0,
1170
+ 0.0,
1171
+ 1.0,
1172
+ 0.0,
1173
+ 1.0,
1174
+ 0.0,
1175
+ 0.0,
1176
+ 1.0,
1177
+ 0.0,
1178
+ 1.0,
1179
+ 1.0,
1180
+ 1.0,
1181
+ 1.0,
1182
+ 1.0,
1183
+ 0.0,
1184
+ 1.0,
1185
+ 0.0,
1186
+ 1.0,
1187
+ 1.0,
1188
+ 1.0,
1189
+ 1.0,
1190
+ 1.0,
1191
+ 1.0,
1192
+ 1.0,
1193
+ 0.0,
1194
+ 1.0,
1195
+ 1.0,
1196
+ 1.0,
1197
+ 1.0,
1198
+ 1.0,
1199
+ 0.0,
1200
+ 1.0,
1201
+ 1.0,
1202
+ 1.0,
1203
+ 0.0,
1204
+ 0.0,
1205
+ 1.0,
1206
+ 1.0,
1207
+ 1.0,
1208
+ 0.0,
1209
+ 1.0,
1210
+ 1.0,
1211
+ 1.0,
1212
+ 1.0,
1213
+ 1.0,
1214
+ 1.0,
1215
+ 0.0,
1216
+ 0.0,
1217
+ 1.0,
1218
+ 1.0,
1219
+ 1.0,
1220
+ 1.0,
1221
+ 1.0,
1222
+ 1.0,
1223
+ 0.0,
1224
+ 0.0,
1225
+ 0.0,
1226
+ 1.0,
1227
+ 1.0,
1228
+ 1.0,
1229
+ 0.0,
1230
+ 1.0,
1231
+ 0.0,
1232
+ 0.0,
1233
+ 0.0,
1234
+ 1.0,
1235
+ 0.0,
1236
+ 0.0,
1237
+ 1.0,
1238
+ 0.0,
1239
+ 0.0,
1240
+ 0.0,
1241
+ 0.0,
1242
+ 1.0,
1243
+ 1.0,
1244
+ 1.0,
1245
+ 1.0,
1246
+ 1.0,
1247
+ 0.0,
1248
+ 1.0,
1249
+ 0.0,
1250
+ 1.0,
1251
+ 1.0,
1252
+ 1.0,
1253
+ 1.0,
1254
+ 0.0,
1255
+ 1.0,
1256
+ 0.0,
1257
+ 1.0,
1258
+ 1.0,
1259
+ 1.0,
1260
+ 1.0,
1261
+ 1.0,
1262
+ 0.0,
1263
+ 1.0,
1264
+ 0.0,
1265
+ 0.0,
1266
+ 1.0,
1267
+ 0.0,
1268
+ 0.0,
1269
+ 1.0,
1270
+ 0.0
1271
+ ],
1272
+ "endpoint_true_t3_mm": {
1273
+ "ci95_high": 148.04044950501992,
1274
+ "ci95_low": 130.49022566219793,
1275
+ "mean": 139.19410413503647
1276
+ },
1277
+ "endpoint_wrong_t3_mm": {
1278
+ "ci95_high": 164.38267045468092,
1279
+ "ci95_low": 143.8837217701599,
1280
+ "mean": 153.92470409721136
1281
+ },
1282
+ "n_episodes": 256,
1283
+ "n_novel_compositions": 64,
1284
+ "novel_pair_discrimination_acc": {
1285
+ "ci95_high": 0.609375,
1286
+ "ci95_low": 0.359375,
1287
+ "mean": 0.484375
1288
+ },
1289
+ "novel_pair_step4_mm": {
1290
+ "ci95_high": 7.899430900998413,
1291
+ "ci95_low": 5.76528804525733,
1292
+ "mean": 6.796592680737376
1293
+ },
1294
+ "policy_cosine_gain": {
1295
+ "ci95_high": 0.024829665957250315,
1296
+ "ci95_low": -0.031018196921399975,
1297
+ "mean": -0.003583221090252664
1298
+ },
1299
+ "policy_cosine_true": 0.16118120129158342,
1300
+ "policy_cosine_wrong": 0.16476442238183608,
1301
+ "step1_error_mm": {
1302
+ "ci95_high": 5.00969171046745,
1303
+ "ci95_low": 4.2749277805094605,
1304
+ "mean": 4.626493820454925
1305
+ },
1306
+ "step1_persistence_mm": 31.911159235984087,
1307
+ "step1_skill_vs_persistence": 0.8550195627102779,
1308
+ "step4_counterfactual_plan_mm": 169.8988118469715,
1309
+ "step4_error_mm": {
1310
+ "ci95_high": 8.249065987113864,
1311
+ "ci95_low": 6.580149448243901,
1312
+ "mean": 7.336826083250344
1313
+ },
1314
+ "step4_mm_vec": [
1315
+ 17.64449119567871,
1316
+ 7.113313674926758,
1317
+ 5.504810333251953,
1318
+ 18.235660552978516,
1319
+ 8.504068374633789,
1320
+ 13.594381332397461,
1321
+ 20.65494155883789,
1322
+ 2.088914632797241,
1323
+ 2.688481092453003,
1324
+ 10.113476753234863,
1325
+ 2.6561594009399414,
1326
+ 8.684379577636719,
1327
+ 8.149361610412598,
1328
+ 84.12359619140625,
1329
+ 11.405562400817871,
1330
+ 5.175888538360596,
1331
+ 13.341816902160645,
1332
+ 3.254366636276245,
1333
+ 10.460737228393555,
1334
+ 2.0569684505462646,
1335
+ 6.295121669769287,
1336
+ 4.076549053192139,
1337
+ 4.577296257019043,
1338
+ 2.362372398376465,
1339
+ 3.3624720573425293,
1340
+ 7.309418678283691,
1341
+ 3.7660715579986572,
1342
+ 13.448225975036621,
1343
+ 5.299437522888184,
1344
+ 4.243538856506348,
1345
+ 1.9432731866836548,
1346
+ 6.805112838745117,
1347
+ 7.5433878898620605,
1348
+ 3.402245283126831,
1349
+ 4.2523627281188965,
1350
+ 2.331644058227539,
1351
+ 11.169013977050781,
1352
+ 9.99180793762207,
1353
+ 3.520632743835449,
1354
+ 3.7570509910583496,
1355
+ 10.276856422424316,
1356
+ 1.6619149446487427,
1357
+ 12.568821907043457,
1358
+ 6.40442419052124,
1359
+ 8.644320487976074,
1360
+ 7.564281463623047,
1361
+ 4.940453052520752,
1362
+ 2.1408653259277344,
1363
+ 9.815386772155762,
1364
+ 2.758178472518921,
1365
+ 9.293109893798828,
1366
+ 4.667240142822266,
1367
+ 3.3404910564422607,
1368
+ 3.1158039569854736,
1369
+ 3.82180118560791,
1370
+ 3.4805970191955566,
1371
+ 6.402080059051514,
1372
+ 8.015181541442871,
1373
+ 6.3438310623168945,
1374
+ 8.944930076599121,
1375
+ 3.1744401454925537,
1376
+ 6.802624225616455,
1377
+ 1.8688279390335083,
1378
+ 1.5109285116195679,
1379
+ 13.4144926071167,
1380
+ 12.80002498626709,
1381
+ 3.8651256561279297,
1382
+ 3.5669591426849365,
1383
+ 8.840657234191895,
1384
+ 3.1004254817962646,
1385
+ 2.061432361602783,
1386
+ 4.572765350341797,
1387
+ 2.3143539428710938,
1388
+ 4.616228103637695,
1389
+ 26.116422653198242,
1390
+ 2.4884135723114014,
1391
+ 2.0787513256073,
1392
+ 11.031004905700684,
1393
+ 7.154238224029541,
1394
+ 5.012524604797363,
1395
+ 12.67503547668457,
1396
+ 9.196083068847656,
1397
+ 4.775420188903809,
1398
+ 11.57768726348877,
1399
+ 2.823132038116455,
1400
+ 4.8557634353637695,
1401
+ 6.6577301025390625,
1402
+ 4.437429428100586,
1403
+ 2.5559449195861816,
1404
+ 6.702149868011475,
1405
+ 8.060351371765137,
1406
+ 3.1223011016845703,
1407
+ 7.266571044921875,
1408
+ 6.385164737701416,
1409
+ 3.9172744750976562,
1410
+ 11.594902038574219,
1411
+ 7.268889904022217,
1412
+ 3.0578203201293945,
1413
+ 13.003729820251465,
1414
+ 1.8130815029144287,
1415
+ 5.264368057250977,
1416
+ 1.8958598375320435,
1417
+ 12.215262413024902,
1418
+ 8.763213157653809,
1419
+ 6.81106424331665,
1420
+ 6.188709259033203,
1421
+ 12.601476669311523,
1422
+ 3.763235092163086,
1423
+ 3.044757843017578,
1424
+ 5.346630573272705,
1425
+ 5.015413761138916,
1426
+ 15.394227027893066,
1427
+ 2.9651455879211426,
1428
+ 13.987813949584961,
1429
+ 5.4913010597229,
1430
+ 21.64535140991211,
1431
+ 16.618671417236328,
1432
+ 3.9308676719665527,
1433
+ 3.9717788696289062,
1434
+ 1.5410685539245605,
1435
+ 4.133262634277344,
1436
+ 16.595863342285156,
1437
+ 5.926461696624756,
1438
+ 6.98364782333374,
1439
+ 7.13236665725708,
1440
+ 1.615167498588562,
1441
+ 4.5967254638671875,
1442
+ 11.031644821166992,
1443
+ 4.799911975860596,
1444
+ 6.833273887634277,
1445
+ 2.128061294555664,
1446
+ 11.10074234008789,
1447
+ 3.834536552429199,
1448
+ 2.865446090698242,
1449
+ 6.303050994873047,
1450
+ 14.998103141784668,
1451
+ 2.714607000350952,
1452
+ 5.930359840393066,
1453
+ 5.571382522583008,
1454
+ 13.891311645507812,
1455
+ 11.464725494384766,
1456
+ 6.700440883636475,
1457
+ 5.185556888580322,
1458
+ 10.576102256774902,
1459
+ 8.593621253967285,
1460
+ 2.618006944656372,
1461
+ 5.472019672393799,
1462
+ 5.3757758140563965,
1463
+ 17.540706634521484,
1464
+ 19.27073860168457,
1465
+ 5.06244421005249,
1466
+ 4.646507263183594,
1467
+ 9.477828979492188,
1468
+ 8.267663955688477,
1469
+ 5.840665340423584,
1470
+ 5.88379430770874,
1471
+ 6.341081142425537,
1472
+ 4.563035011291504,
1473
+ 5.005764484405518,
1474
+ 5.390143871307373,
1475
+ 6.388686180114746,
1476
+ 13.48536205291748,
1477
+ 6.872714996337891,
1478
+ 7.1761345863342285,
1479
+ 8.889355659484863,
1480
+ 6.095846652984619,
1481
+ 3.7229361534118652,
1482
+ 2.6347169876098633,
1483
+ 1.4022562503814697,
1484
+ 6.523616313934326,
1485
+ 6.416252613067627,
1486
+ 6.155904293060303,
1487
+ 8.166077613830566,
1488
+ 5.172173976898193,
1489
+ 2.3831217288970947,
1490
+ 6.663065433502197,
1491
+ 8.629051208496094,
1492
+ 2.528017520904541,
1493
+ 8.67696762084961,
1494
+ 10.862005233764648,
1495
+ 9.107259750366211,
1496
+ 4.9517693519592285,
1497
+ 3.0458712577819824,
1498
+ 2.791069269180298,
1499
+ 2.3366894721984863,
1500
+ 4.624377727508545,
1501
+ 23.124929428100586,
1502
+ 4.148865699768066,
1503
+ 12.142417907714844,
1504
+ 21.138856887817383,
1505
+ 3.7613682746887207,
1506
+ 1.1463967561721802,
1507
+ 3.680763006210327,
1508
+ 17.196792602539062,
1509
+ 10.201006889343262,
1510
+ 6.471467018127441,
1511
+ 6.871340751647949,
1512
+ 5.687144756317139,
1513
+ 11.939123153686523,
1514
+ 7.066553592681885,
1515
+ 15.293546676635742,
1516
+ 11.989534378051758,
1517
+ 3.389455795288086,
1518
+ 1.6947537660598755,
1519
+ 5.20855712890625,
1520
+ 4.23985481262207,
1521
+ 4.6397786140441895,
1522
+ 10.086617469787598,
1523
+ 6.93662691116333,
1524
+ 2.817779302597046,
1525
+ 4.353731632232666,
1526
+ 3.4769363403320312,
1527
+ 16.304773330688477,
1528
+ 4.982401371002197,
1529
+ 10.395968437194824,
1530
+ 2.1550068855285645,
1531
+ 9.078523635864258,
1532
+ 13.1740083694458,
1533
+ 2.2385401725769043,
1534
+ 13.107372283935547,
1535
+ 4.9982147216796875,
1536
+ 4.361330986022949,
1537
+ 3.3339710235595703,
1538
+ 4.580061435699463,
1539
+ 5.5106940269470215,
1540
+ 11.616721153259277,
1541
+ 3.045423746109009,
1542
+ 29.089948654174805,
1543
+ 1.6460047960281372,
1544
+ 11.574164390563965,
1545
+ 2.7432174682617188,
1546
+ 2.3784611225128174,
1547
+ 3.0781784057617188,
1548
+ 7.768765449523926,
1549
+ 16.291637420654297,
1550
+ 1.8099247217178345,
1551
+ 2.1503446102142334,
1552
+ 3.500584363937378,
1553
+ 23.883407592773438,
1554
+ 3.7504889965057373,
1555
+ 6.145341873168945,
1556
+ 8.17447280883789,
1557
+ 4.007771968841553,
1558
+ 3.919954538345337,
1559
+ 2.7191483974456787,
1560
+ 2.8923802375793457,
1561
+ 3.461338996887207,
1562
+ 4.270954132080078,
1563
+ 5.5699849128723145,
1564
+ 3.3417468070983887,
1565
+ 3.4789202213287354,
1566
+ 17.210405349731445,
1567
+ 9.191872596740723,
1568
+ 2.365368366241455,
1569
+ 13.877253532409668,
1570
+ 8.682072639465332
1571
+ ],
1572
+ "step4_persistence_mm": 87.0783047825098,
1573
+ "step4_skill_vs_persistence": 0.915744500291145,
1574
+ "transient_04s_discrimination_acc": {
1575
+ "ci95_high": 0.59765625,
1576
+ "ci95_low": 0.47265625,
1577
+ "mean": 0.53515625
1578
+ },
1579
+ "transient_04s_late_endpoint_mm": {
1580
+ "ci95_high": 104.51799894236028,
1581
+ "ci95_low": 87.76205490408465,
1582
+ "mean": 95.95052932947874
1583
+ }
1584
+ },
1585
+ "spoofing_and_commit_gate": {
1586
+ "always_commit": {
1587
+ "clean_usefulness": 0.7973977695167286,
1588
+ "ece": 0.41710671596229076,
1589
+ "hidden_barrier_probe_rate": 0.0,
1590
+ "mode_accuracy": 0.5604166388511658,
1591
+ "overall_abstention_rate": 0.0,
1592
+ "spoof_detection_rate": 0.0,
1593
+ "spoofed_action_regret": 0.43848979473114014,
1594
+ "unsupported_claim_rate": 1.0
1595
+ },
1596
+ "fixed_threshold": {
1597
+ "clean_usefulness": 0.718091697645601,
1598
+ "ece": 0.27644848526558946,
1599
+ "hidden_barrier_probe_rate": 0.0,
1600
+ "mode_accuracy": 0.6777777671813965,
1601
+ "overall_abstention_rate": 0.32395833333333335,
1602
+ "spoof_detection_rate": 0.0,
1603
+ "spoofed_action_regret": 0.4430941045284271,
1604
+ "unsupported_claim_rate": 0.0
1605
+ },
1606
+ "learned_innovation": {
1607
+ "clean_usefulness": 0.7311028500619579,
1608
+ "ece": 0.017805285493118897,
1609
+ "hidden_barrier_probe_rate": 0.9384057971014492,
1610
+ "mode_accuracy": 0.9368055462837219,
1611
+ "overall_abstention_rate": 0.175,
1612
+ "spoof_detection_rate": 0.9711934156378601,
1613
+ "spoofed_action_regret": 0.30113738775253296,
1614
+ "unsupported_claim_rate": 0.0
1615
+ }
1616
+ },
1617
+ "thin_barrier_and_persistence": {
1618
+ "barrier_auroc_16x16_subsampled": 0.6767578125,
1619
+ "barrier_auroc_64x64": 1.0,
1620
+ "pymunk_puck_barrier_auroc_64": 1.0,
1621
+ "step1_persistence_rmse": 0.04549631103873253,
1622
+ "step1_rmse": 0.03723597526550293,
1623
+ "step1_skill_vs_persistence": 0.1815605613869946,
1624
+ "step4_persistence_rmse": 0.18016260862350464,
1625
+ "step4_rmse": 0.1080051064491272,
1626
+ "step4_skill_vs_persistence": 0.40051319597158364
1627
+ },
1628
+ "throughput_scaling": {
1629
+ "T128_ms": 44.74,
1630
+ "T128_slots_per_s": 137320.8,
1631
+ "T256_ms": 113.98,
1632
+ "T256_slots_per_s": 107804.4,
1633
+ "T32_ms": 9.84,
1634
+ "T32_slots_per_s": 156083.8,
1635
+ "T512_ms": 317.22,
1636
+ "T512_slots_per_s": 77472.1,
1637
+ "streaming_state_mb_T1024": 14.907,
1638
+ "streaming_state_mb_T64": 14.907
1639
+ },
1640
+ "transient_goal_retention": {
1641
+ "continuous_progress": 0.7612097859382629,
1642
+ "continuous_success": {
1643
+ "ci95_high": 0.9791666666666666,
1644
+ "ci95_low": 0.8125,
1645
+ "mean": 0.8958333333333334
1646
+ },
1647
+ "transient_04s_success": {
1648
+ "ci95_high": 1.0,
1649
+ "ci95_low": 0.9375,
1650
+ "mean": 0.9791666666666666
1651
+ },
1652
+ "transient_drop_pp": {
1653
+ "ci95_high": 0.0,
1654
+ "ci95_low": -18.75,
1655
+ "mean": -8.333333333333334
1656
+ },
1657
+ "transient_progress": 0.839072048664093
1658
+ }
1659
+ }
load_checkpoint.py ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """Read the original SLWM NumPy exports without executing pickle data.
2
+
3
+ The format follows scripts/train/x1b_train.py in ValueAlignedAI/slwm.
4
+ See SOURCE_LICENSE for the original source-code license.
5
+ """
6
+
7
+ from __future__ import annotations
8
+
9
+ import hashlib
10
+ import json
11
+ from pathlib import Path
12
+ from typing import Iterator
13
+
14
+ import numpy as np
15
+
16
+
17
+ def verify_sha256(path: str | Path, expected: str) -> None:
18
+ digest = hashlib.sha256()
19
+ with Path(path).open("rb") as stream:
20
+ for chunk in iter(lambda: stream.read(8 * 1024 * 1024), b""):
21
+ digest.update(chunk)
22
+ if digest.hexdigest() != expected:
23
+ raise ValueError(f"SHA-256 mismatch: {Path(path).name}")
24
+
25
+
26
+ def iter_parameters(path: str | Path) -> Iterator[tuple[str, np.ndarray]]:
27
+ """Yield each named parameter as float32, one array at a time."""
28
+ with np.load(path, allow_pickle=False) as archive:
29
+ if "__dtypes_json__" in archive.files:
30
+ manifest = json.loads(str(archive["__dtypes_json__"].item()))
31
+ expected_keys = set(manifest) | {"__dtypes_json__"}
32
+ if set(archive.files) != expected_keys:
33
+ raise ValueError("The BF16 archive and dtype manifest disagree")
34
+ for key, kind in manifest.items():
35
+ array = archive[key]
36
+ if kind == "bf16_u16":
37
+ if array.dtype.kind != "u" or array.dtype.itemsize != 2:
38
+ raise ValueError(f"Expected uint16 BF16 bits for {key}")
39
+ # BF16 uses the upper 16 bits of an IEEE-754 float32 value.
40
+ value = (array.astype(np.uint32) << 16).view(np.float32)
41
+ elif kind == "float32":
42
+ value = array.astype(np.float32)
43
+ else:
44
+ raise ValueError(f"Unsupported stored dtype: {kind}")
45
+ yield key, value
46
+ elif "__quant_manifest__" in archive.files:
47
+ manifest = json.loads(str(archive["__quant_manifest__"].item()))
48
+ expected_keys = set(manifest) | {"__quant_manifest__"}
49
+ expected_keys.update(key + "__scale" for key, kind in manifest.items() if kind == "int8_pc")
50
+ if set(archive.files) != expected_keys:
51
+ raise ValueError("The INT8 archive and quantization manifest disagree")
52
+ for key, kind in manifest.items():
53
+ array = archive[key]
54
+ if kind == "int8_pc":
55
+ if array.dtype != np.int8:
56
+ raise ValueError(f"Expected int8 values for {key}")
57
+ value = array.astype(np.float32) * archive[key + "__scale"].astype(np.float32)
58
+ elif kind == "float16":
59
+ value = array.astype(np.float32)
60
+ else:
61
+ raise ValueError(f"Unsupported quantization kind: {kind}")
62
+ yield key, value
63
+ else:
64
+ raise ValueError("This file is not a recognized SLWM checkpoint")
65
+
66
+
67
+ def load_checkpoint(path: str | Path, *, expected_sha256: str | None = None) -> dict[str, np.ndarray]:
68
+ """Load a NumPy float32 parameter dictionary; optionally verify its SHA-256.
69
+
70
+ This expands both formats to float32. INT8 storage does not imply INT8
71
+ computation or a similarly small in-memory parameter dictionary.
72
+ """
73
+ if expected_sha256 is not None:
74
+ verify_sha256(path, expected_sha256)
75
+ return dict(iter_parameters(path))
76
+
77
+
78
+ def inspect_checkpoint(path: str | Path) -> dict:
79
+ """Check finite decoded parameters and reproduce the source's strict count."""
80
+ buckets = {"adapters": 0, "processor_core": 0, "jepa_and_uncertainty": 0, "policy_and_heads": 0}
81
+ tensor_count = 0
82
+ for key, value in iter_parameters(path):
83
+ if not np.isfinite(value).all():
84
+ raise ValueError(f"Non-finite weights in {key}")
85
+ tensor_count += 1
86
+ if key.startswith(("vis_", "aud_", "prop_", "text_", "act_", "reg_", "slot_", "mod_")):
87
+ bucket = "adapters"
88
+ elif key.startswith("L") or key == "final_norm":
89
+ bucket = "processor_core"
90
+ elif key.startswith(("jepa_", "unc_", "align_")):
91
+ bucket = "jepa_and_uncertainty"
92
+ else:
93
+ bucket = "policy_and_heads"
94
+ buckets[bucket] += int(value.size)
95
+ buckets["strict_total"] = sum(buckets.values())
96
+ return {"parameter_counts": buckets, "tensor_count": tensor_count, "all_finite": True}
portable_int8.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:faaf635499f1ba8f501ecd64911ffa0bf88969f6637d521337d45ee4bf1ea36f
3
+ size 918039628
receipt.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "arch": "slwm_x1b",
3
+ "backend": "tpu",
4
+ "batch_size": 12,
5
+ "config_sha256": "cca9e2ca4e9ccde614b3261a59d939953218168d70af3bae213952652c02087a",
6
+ "eval_summary": {
7
+ "assoc_T1024": 1.0,
8
+ "assoc_T512": 1.0,
9
+ "av_r1": 0.921875,
10
+ "barrier_auroc_64": 1.0,
11
+ "continuous_success": 0.8958333333333334,
12
+ "g8_shooting_success": 0.23958333333333334,
13
+ "goal_switch_mpc": 1.0,
14
+ "mpc_success": 0.75,
15
+ "persistence_skill": 0.1815605613869946,
16
+ "real_robot_cmd_disc_acc": 0.55859375,
17
+ "real_robot_plan_gain_mm": 162.56198576372117,
18
+ "real_robot_step4_mm": 7.336826083250344,
19
+ "spoof_regret_always": 0.43848979473114014,
20
+ "spoof_regret_learned": 0.30113738775253296,
21
+ "step4_persistence_skill": 0.40051319597158364,
22
+ "sustained_transient_mpc": 1.0,
23
+ "thin_barrier_mpc": 0.9375,
24
+ "transient_04s_success": 0.9791666666666666,
25
+ "transient_drop_pp": -8.333333333333334
26
+ },
27
+ "eval_wall_s": 548.05,
28
+ "objective": "residual_ema_jepa_sigreg",
29
+ "parameter_counts": {
30
+ "adapters": 19033728,
31
+ "jepa_and_uncertainty": 16529666,
32
+ "policy_and_heads": 18272296,
33
+ "processor_core": 954618720,
34
+ "strict_total": 1008454410
35
+ },
36
+ "portable_int8_bytes": 918039628,
37
+ "portable_int8_sha256": "faaf635499f1ba8f501ecd64911ffa0bf88969f6637d521337d45ee4bf1ea36f",
38
+ "profile": "x1b",
39
+ "seed": 2001,
40
+ "seq_len": 32,
41
+ "steps": 5600,
42
+ "tag": "x1b-w5-slwm-1b-s2001",
43
+ "train_wall_s": 2666.11,
44
+ "weights_bf16_bytes": 1602108632,
45
+ "weights_bf16_sha256": "f56b803857a09e7e77e1bb501efdc8e47bcacb2230cf18a1ab44e1acca591f51"
46
+ }
release_manifest.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "repo_id": "valueai-org/SLWM-X1B",
3
+ "title": "SLWM-X1B \u2014 Wave 5",
4
+ "wave": 5,
5
+ "run_tag": "x1b-w5-slwm-1b-s2001",
6
+ "seed": 2001,
7
+ "strict_total_parameters": 1008454410,
8
+ "source_repository": "https://github.com/ValueAlignedAI/slwm",
9
+ "source_reference_commit": "cbb2b38794359adfefe07b19e9b7caa113b25b47",
10
+ "training_source_commit": null,
11
+ "manuscript_version": "0.3",
12
+ "original_artifacts": {
13
+ "config.json": {
14
+ "bytes": 1018,
15
+ "sha256": "2e31e53c6431657da058c6293d3c7b948dfb5b221fe5c4fde1c08ea8c5bd2c00"
16
+ },
17
+ "receipt.json": {
18
+ "bytes": 1566,
19
+ "sha256": "f495c6897d36ececd3f85fad64b5aa7283878e8002a4652bdf0831d584d21347"
20
+ },
21
+ "eval_metrics.json": {
22
+ "bytes": 32177,
23
+ "sha256": "ef8878a15da85fd1e14724e6ad50b6d0a6e6c05f1e1aa673c910f44ff576771b"
24
+ },
25
+ "train.jsonl": {
26
+ "bytes": 41191,
27
+ "sha256": "acb8668b83c886fa6fda8797721ee746cd8552c7def3ab940f8625118e6c084b"
28
+ },
29
+ "weights_bf16.npz": {
30
+ "bytes": 1602108632,
31
+ "sha256": "f56b803857a09e7e77e1bb501efdc8e47bcacb2230cf18a1ab44e1acca591f51"
32
+ },
33
+ "portable_int8.npz": {
34
+ "bytes": 918039628,
35
+ "sha256": "faaf635499f1ba8f501ecd64911ffa0bf88969f6637d521337d45ee4bf1ea36f"
36
+ }
37
+ },
38
+ "validation": {
39
+ "weights_bf16.npz": {
40
+ "parameter_counts": {
41
+ "adapters": 19033728,
42
+ "processor_core": 954618720,
43
+ "jepa_and_uncertainty": 16529666,
44
+ "policy_and_heads": 18272296,
45
+ "strict_total": 1008454410
46
+ },
47
+ "tensor_count": 349,
48
+ "all_finite": true
49
+ },
50
+ "portable_int8.npz": {
51
+ "parameter_counts": {
52
+ "adapters": 19033728,
53
+ "processor_core": 954618720,
54
+ "jepa_and_uncertainty": 16529666,
55
+ "policy_and_heads": 18272296,
56
+ "strict_total": 1008454410
57
+ },
58
+ "tensor_count": 349,
59
+ "all_finite": true
60
+ }
61
+ },
62
+ "metrics_scope": "Original single-seed receipt; no new model-quality evaluation performed."
63
+ }
requirements.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ numpy>=2.0
2
+ huggingface_hub>=1.0
train.jsonl ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1, "goal_steps": 5600, "elapsed_s": 109.64, "steps_per_s": null, "commit_ce": 1.21902, "contact_bce": 1.20491, "goal_mse": 0.41049, "grad_norm": 374.68457, "jepa": 0.00034, "loss": 21.62898, "lr": 0.0, "pr": 0.01059, "recon_mse": 1.08657, "sigreg": 0.56597, "state_mse": 0.0278, "step4_mse": 0.05606, "text_ce": 8.6135}
2
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 50, "goal_steps": 5600, "elapsed_s": 132.04, "steps_per_s": null, "commit_ce": 0.56646, "contact_bce": 0.75367, "goal_mse": 0.24324, "grad_norm": 54.34883, "jepa": 0.0011, "loss": 7.01363, "lr": 5e-05, "pr": 0.0083, "recon_mse": 0.38214, "sigreg": 0.63363, "state_mse": 0.01446, "step4_mse": 0.03403, "text_ce": 2.58051}
3
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 100, "goal_steps": 5600, "elapsed_s": 154.89, "steps_per_s": 2.188, "commit_ce": 0.56391, "contact_bce": 0.63586, "goal_mse": 0.10404, "grad_norm": 42.37888, "jepa": 0.00067, "loss": 4.84901, "lr": 0.0001, "pr": 0.01197, "recon_mse": 0.12228, "sigreg": 0.57621, "state_mse": 0.00612, "step4_mse": 0.01776, "text_ce": 1.85565}
4
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 150, "goal_steps": 5600, "elapsed_s": 177.71, "steps_per_s": 2.191, "commit_ce": 0.34909, "contact_bce": 0.47086, "goal_mse": 0.20734, "grad_norm": 60.57965, "jepa": 0.0005, "loss": 6.6353, "lr": 0.00015, "pr": 0.01975, "recon_mse": 0.15913, "sigreg": 0.50881, "state_mse": 0.00536, "step4_mse": 0.01657, "text_ce": 1.13917}
5
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 200, "goal_steps": 5600, "elapsed_s": 200.55, "steps_per_s": 2.189, "commit_ce": 0.11831, "contact_bce": 0.43367, "goal_mse": 0.14101, "grad_norm": 44.83906, "jepa": 0.00057, "loss": 3.16793, "lr": 0.0002, "pr": 0.01458, "recon_mse": 0.19191, "sigreg": 0.5355, "state_mse": 0.005, "step4_mse": 0.01099, "text_ce": 0.95124}
6
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 250, "goal_steps": 5600, "elapsed_s": 223.34, "steps_per_s": 2.194, "commit_ce": 0.19768, "contact_bce": 0.42928, "goal_mse": 0.05881, "grad_norm": 35.13424, "jepa": 0.0004, "loss": 5.23115, "lr": 0.00024, "pr": 0.02336, "recon_mse": 0.13716, "sigreg": 0.4432, "state_mse": 0.00437, "step4_mse": 0.01249, "text_ce": 1.08697}
7
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 300, "goal_steps": 5600, "elapsed_s": 246.15, "steps_per_s": 2.192, "commit_ce": 0.3557, "contact_bce": 0.43521, "goal_mse": 0.05301, "grad_norm": 29.94538, "jepa": 0.00019, "loss": 2.27445, "lr": 0.00024, "pr": 0.0193, "recon_mse": 0.13942, "sigreg": 0.4716, "state_mse": 0.00574, "step4_mse": 0.01331, "text_ce": 0.33836}
8
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 350, "goal_steps": 5600, "elapsed_s": 268.97, "steps_per_s": 2.191, "commit_ce": 0.47756, "contact_bce": 0.43741, "goal_mse": 0.03077, "grad_norm": 22.26558, "jepa": 0.00026, "loss": 2.14899, "lr": 0.00024, "pr": 0.03173, "recon_mse": 0.13177, "sigreg": 0.39496, "state_mse": 0.00547, "step4_mse": 0.01362, "text_ce": 0.40528}
9
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 400, "goal_steps": 5600, "elapsed_s": 291.77, "steps_per_s": 2.193, "commit_ce": 0.32163, "contact_bce": 0.35276, "goal_mse": 0.06011, "grad_norm": 15.88012, "jepa": 0.00023, "loss": 4.59528, "lr": 0.00024, "pr": 0.03337, "recon_mse": 0.08211, "sigreg": 0.37528, "state_mse": 0.00234, "step4_mse": 0.00514, "text_ce": 1.26127}
10
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 450, "goal_steps": 5600, "elapsed_s": 314.56, "steps_per_s": 2.195, "commit_ce": 0.30515, "contact_bce": 0.39411, "goal_mse": 0.07252, "grad_norm": 35.79497, "jepa": 0.00021, "loss": 5.62526, "lr": 0.00024, "pr": 0.03102, "recon_mse": 0.12078, "sigreg": 0.41049, "state_mse": 0.00403, "step4_mse": 0.00843, "text_ce": 0.59892}
11
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 500, "goal_steps": 5600, "elapsed_s": 337.35, "steps_per_s": 2.193, "commit_ce": 0.1317, "contact_bce": 0.25975, "goal_mse": 0.14261, "grad_norm": 42.9335, "jepa": 0.00034, "loss": 3.87415, "lr": 0.00024, "pr": 0.03383, "recon_mse": 0.10465, "sigreg": 0.37736, "state_mse": 0.00232, "step4_mse": 0.00575, "text_ce": 0.01126}
12
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 550, "goal_steps": 5600, "elapsed_s": 360.18, "steps_per_s": 2.19, "commit_ce": 0.14633, "contact_bce": 0.34638, "goal_mse": 0.07431, "grad_norm": 18.85131, "jepa": 0.00024, "loss": 3.95097, "lr": 0.00024, "pr": 0.0322, "recon_mse": 0.12552, "sigreg": 0.36931, "state_mse": 0.0038, "step4_mse": 0.00793, "text_ce": 0.47518}
13
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 600, "goal_steps": 5600, "elapsed_s": 383.03, "steps_per_s": 2.188, "commit_ce": 0.29575, "contact_bce": 0.36661, "goal_mse": 0.05874, "grad_norm": 27.2921, "jepa": 0.00023, "loss": 1.66785, "lr": 0.00024, "pr": 0.03418, "recon_mse": 0.10487, "sigreg": 0.35033, "state_mse": 0.00619, "step4_mse": 0.0089, "text_ce": 0.14956}
14
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 650, "goal_steps": 5600, "elapsed_s": 405.83, "steps_per_s": 2.193, "commit_ce": 0.25442, "contact_bce": 0.35349, "goal_mse": 0.06738, "grad_norm": 30.04128, "jepa": 0.00017, "loss": 1.35552, "lr": 0.00024, "pr": 0.03225, "recon_mse": 0.13435, "sigreg": 0.37009, "state_mse": 0.00407, "step4_mse": 0.00809, "text_ce": 0.03715}
15
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 700, "goal_steps": 5600, "elapsed_s": 428.62, "steps_per_s": 2.194, "commit_ce": 0.28356, "contact_bce": 0.37733, "goal_mse": 0.08812, "grad_norm": 8.4238, "jepa": 0.00026, "loss": 3.80779, "lr": 0.00024, "pr": 0.03605, "recon_mse": 0.11922, "sigreg": 0.34015, "state_mse": 0.00273, "step4_mse": 0.00581, "text_ce": 0.02609}
16
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 750, "goal_steps": 5600, "elapsed_s": 451.42, "steps_per_s": 2.193, "commit_ce": 0.38319, "contact_bce": 0.4661, "goal_mse": 0.06248, "grad_norm": 23.48066, "jepa": 0.00019, "loss": 3.99294, "lr": 0.00024, "pr": 0.03124, "recon_mse": 0.09193, "sigreg": 0.37076, "state_mse": 0.00331, "step4_mse": 0.00845, "text_ce": 0.37064}
17
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 800, "goal_steps": 5600, "elapsed_s": 474.23, "steps_per_s": 2.193, "commit_ce": 0.15348, "contact_bce": 0.32414, "goal_mse": 0.05259, "grad_norm": 14.41723, "jepa": 0.00027, "loss": 3.94342, "lr": 0.00023, "pr": 0.03674, "recon_mse": 0.09845, "sigreg": 0.32352, "state_mse": 0.00359, "step4_mse": 0.00668, "text_ce": 0.2575}
18
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 850, "goal_steps": 5600, "elapsed_s": 497.02, "steps_per_s": 2.193, "commit_ce": 0.28859, "contact_bce": 0.37073, "goal_mse": 0.05409, "grad_norm": 26.97801, "jepa": 0.0002, "loss": 1.20915, "lr": 0.00023, "pr": 0.03435, "recon_mse": 0.10854, "sigreg": 0.33, "state_mse": 0.00246, "step4_mse": 0.00537, "text_ce": 0.01323}
19
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 900, "goal_steps": 5600, "elapsed_s": 519.87, "steps_per_s": 2.189, "commit_ce": 0.14624, "contact_bce": 0.306, "goal_mse": 0.04229, "grad_norm": 19.69111, "jepa": 0.00024, "loss": 1.59051, "lr": 0.00023, "pr": 0.03511, "recon_mse": 0.13369, "sigreg": 0.3371, "state_mse": 0.00309, "step4_mse": 0.00533, "text_ce": 0.48072}
20
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 950, "goal_steps": 5600, "elapsed_s": 542.72, "steps_per_s": 2.188, "commit_ce": 0.33454, "contact_bce": 0.28891, "goal_mse": 0.05134, "grad_norm": 22.30033, "jepa": 0.0002, "loss": 3.5038, "lr": 0.00023, "pr": 0.03801, "recon_mse": 0.09572, "sigreg": 0.32087, "state_mse": 0.00267, "step4_mse": 0.00595, "text_ce": 0.19186}
21
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1000, "goal_steps": 5600, "elapsed_s": 565.56, "steps_per_s": 2.188, "commit_ce": 0.29335, "contact_bce": 0.30059, "goal_mse": 0.05505, "grad_norm": 18.95374, "jepa": 0.00045, "loss": 3.07447, "lr": 0.00023, "pr": 0.03608, "recon_mse": 0.12669, "sigreg": 0.35108, "state_mse": 0.00377, "step4_mse": 0.00653, "text_ce": 0.00164}
22
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1050, "goal_steps": 5600, "elapsed_s": 588.42, "steps_per_s": 2.188, "commit_ce": 0.30773, "contact_bce": 0.37645, "goal_mse": 0.04581, "grad_norm": 9.09284, "jepa": 0.00021, "loss": 1.1513, "lr": 0.00023, "pr": 0.0392, "recon_mse": 0.07469, "sigreg": 0.30085, "state_mse": 0.0025, "step4_mse": 0.00589, "text_ce": 0.0005}
23
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1100, "goal_steps": 5600, "elapsed_s": 611.26, "steps_per_s": 2.188, "commit_ce": 0.25524, "contact_bce": 0.28513, "goal_mse": 0.04953, "grad_norm": 21.36717, "jepa": 0.00022, "loss": 1.06997, "lr": 0.00023, "pr": 0.03519, "recon_mse": 0.12265, "sigreg": 0.3376, "state_mse": 0.00192, "step4_mse": 0.0039, "text_ce": 0.00123}
24
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1150, "goal_steps": 5600, "elapsed_s": 634.11, "steps_per_s": 2.188, "commit_ce": 0.17365, "contact_bce": 0.3559, "goal_mse": 0.04056, "grad_norm": 18.09151, "jepa": 0.00033, "loss": 3.30771, "lr": 0.00022, "pr": 0.05102, "recon_mse": 0.07175, "sigreg": 0.25764, "state_mse": 0.00178, "step4_mse": 0.00361, "text_ce": 0.00763}
25
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1200, "goal_steps": 5600, "elapsed_s": 656.97, "steps_per_s": 2.188, "commit_ce": 0.20985, "contact_bce": 0.31538, "goal_mse": 0.03623, "grad_norm": 8.68808, "jepa": 0.00025, "loss": 3.65166, "lr": 0.00022, "pr": 0.04327, "recon_mse": 0.09509, "sigreg": 0.28195, "state_mse": 0.00189, "step4_mse": 0.00421, "text_ce": 0.00088}
26
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1250, "goal_steps": 5600, "elapsed_s": 679.77, "steps_per_s": 2.193, "commit_ce": 0.14581, "contact_bce": 0.25915, "goal_mse": 0.03733, "grad_norm": 5.68357, "jepa": 0.0004, "loss": 3.18876, "lr": 0.00022, "pr": 0.05108, "recon_mse": 0.09164, "sigreg": 0.25217, "state_mse": 0.00171, "step4_mse": 0.00512, "text_ce": 0.00033}
27
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1300, "goal_steps": 5600, "elapsed_s": 702.59, "steps_per_s": 2.191, "commit_ce": 0.13276, "contact_bce": 0.23684, "goal_mse": 0.06494, "grad_norm": 19.34422, "jepa": 0.00031, "loss": 2.92777, "lr": 0.00022, "pr": 0.05433, "recon_mse": 0.05447, "sigreg": 0.24077, "state_mse": 0.00208, "step4_mse": 0.00554, "text_ce": 0.00029}
28
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1350, "goal_steps": 5600, "elapsed_s": 725.38, "steps_per_s": 2.194, "commit_ce": 0.09003, "contact_bce": 0.31697, "goal_mse": 0.0544, "grad_norm": 8.89243, "jepa": 0.00044, "loss": 3.48054, "lr": 0.00022, "pr": 0.05072, "recon_mse": 0.08905, "sigreg": 0.2564, "state_mse": 0.00137, "step4_mse": 0.00322, "text_ce": 0.00035}
29
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1400, "goal_steps": 5600, "elapsed_s": 748.18, "steps_per_s": 2.194, "commit_ce": 0.12108, "contact_bce": 0.27409, "goal_mse": 0.04216, "grad_norm": 21.78035, "jepa": 0.0004, "loss": 0.80661, "lr": 0.00022, "pr": 0.0537, "recon_mse": 0.09488, "sigreg": 0.23823, "state_mse": 0.00178, "step4_mse": 0.00424, "text_ce": 0.00128}
30
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1450, "goal_steps": 5600, "elapsed_s": 770.96, "steps_per_s": 2.194, "commit_ce": 0.17463, "contact_bce": 0.29271, "goal_mse": 0.12148, "grad_norm": 38.88994, "jepa": 0.0004, "loss": 1.18267, "lr": 0.00021, "pr": 0.05594, "recon_mse": 0.12662, "sigreg": 0.23206, "state_mse": 0.0013, "step4_mse": 0.00361, "text_ce": 0.00044}
31
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1500, "goal_steps": 5600, "elapsed_s": 793.76, "steps_per_s": 2.193, "commit_ce": 0.16236, "contact_bce": 0.28044, "goal_mse": 0.06135, "grad_norm": 14.64709, "jepa": 0.00037, "loss": 2.97745, "lr": 0.00021, "pr": 0.05528, "recon_mse": 0.06685, "sigreg": 0.24332, "state_mse": 0.00134, "step4_mse": 0.00298, "text_ce": 0.00949}
32
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1550, "goal_steps": 5600, "elapsed_s": 816.55, "steps_per_s": 2.194, "commit_ce": 0.37481, "contact_bce": 0.33684, "goal_mse": 0.03877, "grad_norm": 8.69105, "jepa": 0.00023, "loss": 0.98244, "lr": 0.00021, "pr": 0.04289, "recon_mse": 0.06791, "sigreg": 0.28091, "state_mse": 0.00143, "step4_mse": 0.00303, "text_ce": 0.04848}
33
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1600, "goal_steps": 5600, "elapsed_s": 839.38, "steps_per_s": 2.191, "commit_ce": 0.32629, "contact_bce": 0.21895, "goal_mse": 0.0529, "grad_norm": 18.70959, "jepa": 0.00029, "loss": 3.4233, "lr": 0.00021, "pr": 0.05604, "recon_mse": 0.07809, "sigreg": 0.23267, "state_mse": 0.00143, "step4_mse": 0.00302, "text_ce": 0.00179}
34
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1650, "goal_steps": 5600, "elapsed_s": 862.22, "steps_per_s": 2.189, "commit_ce": 0.20218, "contact_bce": 0.30882, "goal_mse": 0.03524, "grad_norm": 4.93382, "jepa": 0.00027, "loss": 0.7602, "lr": 0.00021, "pr": 0.05288, "recon_mse": 0.09116, "sigreg": 0.23608, "state_mse": 0.00151, "step4_mse": 0.00529, "text_ce": 0.00016}
35
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1700, "goal_steps": 5600, "elapsed_s": 885.01, "steps_per_s": 2.194, "commit_ce": 0.17811, "contact_bce": 0.22355, "goal_mse": 0.0451, "grad_norm": 10.484, "jepa": 0.00043, "loss": 0.69154, "lr": 0.0002, "pr": 0.052, "recon_mse": 0.08811, "sigreg": 0.24014, "state_mse": 0.00149, "step4_mse": 0.00332, "text_ce": 0.02498}
36
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1750, "goal_steps": 5600, "elapsed_s": 907.8, "steps_per_s": 2.194, "commit_ce": 0.23642, "contact_bce": 0.2949, "goal_mse": 0.05915, "grad_norm": 9.13847, "jepa": 0.00039, "loss": 3.035, "lr": 0.0002, "pr": 0.0592, "recon_mse": 0.04755, "sigreg": 0.22074, "state_mse": 0.00144, "step4_mse": 0.00298, "text_ce": 0.00021}
37
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1800, "goal_steps": 5600, "elapsed_s": 930.59, "steps_per_s": 2.194, "commit_ce": 0.13788, "contact_bce": 0.26414, "goal_mse": 0.03611, "grad_norm": 16.50298, "jepa": 0.00033, "loss": 0.76672, "lr": 0.0002, "pr": 0.06023, "recon_mse": 0.09868, "sigreg": 0.20855, "state_mse": 0.00199, "step4_mse": 0.00396, "text_ce": 0.0002}
38
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1850, "goal_steps": 5600, "elapsed_s": 953.39, "steps_per_s": 2.193, "commit_ce": 0.14957, "contact_bce": 0.24892, "goal_mse": 0.06746, "grad_norm": 9.45729, "jepa": 0.00027, "loss": 2.88644, "lr": 0.0002, "pr": 0.04999, "recon_mse": 0.06147, "sigreg": 0.2427, "state_mse": 0.00149, "step4_mse": 0.00459, "text_ce": 0.00515}
39
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1900, "goal_steps": 5600, "elapsed_s": 976.19, "steps_per_s": 2.193, "commit_ce": 0.21591, "contact_bce": 0.53916, "goal_mse": 0.01338, "grad_norm": 8.44576, "jepa": 0.00038, "loss": 3.44385, "lr": 0.00019, "pr": 0.06396, "recon_mse": 0.08489, "sigreg": 0.19688, "state_mse": 0.00103, "step4_mse": 0.00292, "text_ce": 0.05989}
40
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 1950, "goal_steps": 5600, "elapsed_s": 999.03, "steps_per_s": 2.188, "commit_ce": 0.1126, "contact_bce": 0.30635, "goal_mse": 0.0251, "grad_norm": 12.27227, "jepa": 0.00028, "loss": 0.77211, "lr": 0.00019, "pr": 0.05058, "recon_mse": 0.10735, "sigreg": 0.2313, "state_mse": 0.00156, "step4_mse": 0.00352, "text_ce": 0.00077}
41
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2000, "goal_steps": 5600, "elapsed_s": 1021.83, "steps_per_s": 2.193, "commit_ce": 0.15687, "contact_bce": 0.37802, "goal_mse": 0.00862, "grad_norm": 6.34391, "jepa": 0.00033, "loss": 0.7976, "lr": 0.00019, "pr": 0.06003, "recon_mse": 0.08616, "sigreg": 0.20978, "state_mse": 0.00146, "step4_mse": 0.00344, "text_ce": 0.00026}
42
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2050, "goal_steps": 5600, "elapsed_s": 1044.62, "steps_per_s": 2.194, "commit_ce": 0.13606, "contact_bce": 0.25304, "goal_mse": 0.04862, "grad_norm": 9.43029, "jepa": 0.00037, "loss": 2.22559, "lr": 0.00018, "pr": 0.06276, "recon_mse": 0.06704, "sigreg": 0.19627, "state_mse": 0.00165, "step4_mse": 0.00401, "text_ce": 0.00019}
43
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2100, "goal_steps": 5600, "elapsed_s": 1067.45, "steps_per_s": 2.19, "commit_ce": 0.15471, "contact_bce": 0.25399, "goal_mse": 0.01826, "grad_norm": 7.08862, "jepa": 0.00027, "loss": 2.77362, "lr": 0.00018, "pr": 0.05373, "recon_mse": 0.04152, "sigreg": 0.22918, "state_mse": 0.001, "step4_mse": 0.00365, "text_ce": 0.00017}
44
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2150, "goal_steps": 5600, "elapsed_s": 1090.32, "steps_per_s": 2.186, "commit_ce": 0.31511, "contact_bce": 0.238, "goal_mse": 0.02913, "grad_norm": 10.82196, "jepa": 0.00033, "loss": 0.77979, "lr": 0.00018, "pr": 0.05933, "recon_mse": 0.10892, "sigreg": 0.20815, "state_mse": 0.00118, "step4_mse": 0.0028, "text_ce": 0.00094}
45
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2200, "goal_steps": 5600, "elapsed_s": 1113.19, "steps_per_s": 2.187, "commit_ce": 0.1992, "contact_bce": 0.33768, "goal_mse": 0.01496, "grad_norm": 7.33129, "jepa": 0.00036, "loss": 2.52788, "lr": 0.00018, "pr": 0.06849, "recon_mse": 0.04819, "sigreg": 0.19533, "state_mse": 0.00101, "step4_mse": 0.00283, "text_ce": 0.00016}
46
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2250, "goal_steps": 5600, "elapsed_s": 1136.05, "steps_per_s": 2.187, "commit_ce": 0.16165, "contact_bce": 0.27751, "goal_mse": 0.03461, "grad_norm": 9.46979, "jepa": 0.0004, "loss": 3.04662, "lr": 0.00017, "pr": 0.06794, "recon_mse": 0.09452, "sigreg": 0.19446, "state_mse": 0.00155, "step4_mse": 0.00344, "text_ce": 0.00012}
47
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2300, "goal_steps": 5600, "elapsed_s": 1158.95, "steps_per_s": 2.184, "commit_ce": 0.28327, "contact_bce": 0.3255, "goal_mse": 0.02677, "grad_norm": 6.60747, "jepa": 0.00026, "loss": 0.78954, "lr": 0.00017, "pr": 0.05724, "recon_mse": 0.09549, "sigreg": 0.21491, "state_mse": 0.0014, "step4_mse": 0.00271, "text_ce": 0.0006}
48
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2350, "goal_steps": 5600, "elapsed_s": 1181.81, "steps_per_s": 2.187, "commit_ce": 0.30594, "contact_bce": 0.23047, "goal_mse": 0.01443, "grad_norm": 3.17185, "jepa": 0.00026, "loss": 0.59053, "lr": 0.00017, "pr": 0.05206, "recon_mse": 0.08172, "sigreg": 0.22274, "state_mse": 0.00082, "step4_mse": 0.00216, "text_ce": 0.00027}
49
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2400, "goal_steps": 5600, "elapsed_s": 1204.67, "steps_per_s": 2.187, "commit_ce": 0.21778, "contact_bce": 0.29338, "goal_mse": 0.02152, "grad_norm": 9.03307, "jepa": 0.00031, "loss": 0.64063, "lr": 0.00016, "pr": 0.06438, "recon_mse": 0.09165, "sigreg": 0.19462, "state_mse": 0.00087, "step4_mse": 0.00202, "text_ce": 0.00012}
50
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2450, "goal_steps": 5600, "elapsed_s": 1227.51, "steps_per_s": 2.189, "commit_ce": 0.09087, "contact_bce": 0.24636, "goal_mse": 0.02443, "grad_norm": 6.10774, "jepa": 0.00032, "loss": 0.57887, "lr": 0.00016, "pr": 0.06448, "recon_mse": 0.07463, "sigreg": 0.19002, "state_mse": 0.00142, "step4_mse": 0.00297, "text_ce": 0.00206}
51
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2500, "goal_steps": 5600, "elapsed_s": 1250.36, "steps_per_s": 2.189, "commit_ce": 0.32403, "contact_bce": 0.24524, "goal_mse": 0.01549, "grad_norm": 6.34463, "jepa": 0.00024, "loss": 0.67632, "lr": 0.00016, "pr": 0.05739, "recon_mse": 0.06087, "sigreg": 0.20441, "state_mse": 0.00103, "step4_mse": 0.00343, "text_ce": 0.00019}
52
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2550, "goal_steps": 5600, "elapsed_s": 1273.19, "steps_per_s": 2.19, "commit_ce": 0.11517, "contact_bce": 0.25777, "goal_mse": 0.01672, "grad_norm": 10.11508, "jepa": 0.00026, "loss": 0.54269, "lr": 0.00016, "pr": 0.06144, "recon_mse": 0.05204, "sigreg": 0.2039, "state_mse": 0.00115, "step4_mse": 0.0026, "text_ce": 0.0001}
53
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2600, "goal_steps": 5600, "elapsed_s": 1296.02, "steps_per_s": 2.19, "commit_ce": 0.1276, "contact_bce": 0.29004, "goal_mse": 0.01445, "grad_norm": 10.10156, "jepa": 0.00034, "loss": 0.64483, "lr": 0.00015, "pr": 0.0758, "recon_mse": 0.09234, "sigreg": 0.16619, "state_mse": 0.00138, "step4_mse": 0.00322, "text_ce": 9e-05}
54
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2650, "goal_steps": 5600, "elapsed_s": 1318.83, "steps_per_s": 2.191, "commit_ce": 0.14669, "contact_bce": 0.18681, "goal_mse": 0.02314, "grad_norm": 11.11628, "jepa": 0.00029, "loss": 1.02732, "lr": 0.00015, "pr": 0.07061, "recon_mse": 0.06485, "sigreg": 0.17882, "state_mse": 0.00116, "step4_mse": 0.00346, "text_ce": 0.0001}
55
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2700, "goal_steps": 5600, "elapsed_s": 1341.67, "steps_per_s": 2.189, "commit_ce": 0.14781, "contact_bce": 0.22816, "goal_mse": 0.02246, "grad_norm": 7.27771, "jepa": 0.00032, "loss": 0.85702, "lr": 0.00015, "pr": 0.07962, "recon_mse": 0.07035, "sigreg": 0.15979, "state_mse": 0.00173, "step4_mse": 0.00429, "text_ce": 0.00013}
56
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2750, "goal_steps": 5600, "elapsed_s": 1364.5, "steps_per_s": 2.19, "commit_ce": 0.22498, "contact_bce": 0.21745, "goal_mse": 0.01095, "grad_norm": 7.92109, "jepa": 0.00033, "loss": 0.43991, "lr": 0.00014, "pr": 0.07263, "recon_mse": 0.07846, "sigreg": 0.17961, "state_mse": 0.00062, "step4_mse": 0.00132, "text_ce": 8e-05}
57
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2800, "goal_steps": 5600, "elapsed_s": 1387.35, "steps_per_s": 2.188, "commit_ce": 0.20676, "contact_bce": 0.23598, "goal_mse": 0.05493, "grad_norm": 12.30188, "jepa": 0.0003, "loss": 0.64528, "lr": 0.00014, "pr": 0.07633, "recon_mse": 0.05444, "sigreg": 0.15936, "state_mse": 0.00089, "step4_mse": 0.00265, "text_ce": 7e-05}
58
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2850, "goal_steps": 5600, "elapsed_s": 1410.2, "steps_per_s": 2.188, "commit_ce": 0.12557, "contact_bce": 0.27609, "goal_mse": 0.02273, "grad_norm": 6.42753, "jepa": 0.00034, "loss": 1.3633, "lr": 0.00014, "pr": 0.07788, "recon_mse": 0.05266, "sigreg": 0.17397, "state_mse": 0.00124, "step4_mse": 0.00267, "text_ce": 0.00012}
59
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2900, "goal_steps": 5600, "elapsed_s": 1433.08, "steps_per_s": 2.186, "commit_ce": 0.22659, "contact_bce": 0.19323, "goal_mse": 0.02016, "grad_norm": 7.91377, "jepa": 0.00033, "loss": 0.93705, "lr": 0.00013, "pr": 0.07999, "recon_mse": 0.05989, "sigreg": 0.15268, "state_mse": 0.00115, "step4_mse": 0.00254, "text_ce": 0.00011}
60
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 2950, "goal_steps": 5600, "elapsed_s": 1455.93, "steps_per_s": 2.188, "commit_ce": 0.29867, "contact_bce": 0.24297, "goal_mse": 0.01177, "grad_norm": 7.29544, "jepa": 0.00044, "loss": 1.37409, "lr": 0.00013, "pr": 0.08034, "recon_mse": 0.06914, "sigreg": 0.16506, "state_mse": 0.00119, "step4_mse": 0.00242, "text_ce": 0.00016}
61
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3000, "goal_steps": 5600, "elapsed_s": 1478.78, "steps_per_s": 2.188, "commit_ce": 0.20021, "contact_bce": 0.29823, "goal_mse": 0.02757, "grad_norm": 9.52911, "jepa": 0.00032, "loss": 0.74601, "lr": 0.00013, "pr": 0.07256, "recon_mse": 0.06671, "sigreg": 0.17397, "state_mse": 0.00156, "step4_mse": 0.00542, "text_ce": 8e-05}
62
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3050, "goal_steps": 5600, "elapsed_s": 1501.65, "steps_per_s": 2.187, "commit_ce": 0.16522, "contact_bce": 0.25938, "goal_mse": 0.01456, "grad_norm": 3.36131, "jepa": 0.00032, "loss": 0.7105, "lr": 0.00012, "pr": 0.08639, "recon_mse": 0.04036, "sigreg": 0.1449, "state_mse": 0.00094, "step4_mse": 0.00192, "text_ce": 7e-05}
63
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3100, "goal_steps": 5600, "elapsed_s": 1524.49, "steps_per_s": 2.188, "commit_ce": 0.19396, "contact_bce": 0.23868, "goal_mse": 0.0157, "grad_norm": 9.38671, "jepa": 0.00028, "loss": 1.87213, "lr": 0.00012, "pr": 0.07197, "recon_mse": 0.04983, "sigreg": 0.17852, "state_mse": 0.00162, "step4_mse": 0.0042, "text_ce": 5e-05}
64
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3150, "goal_steps": 5600, "elapsed_s": 1547.32, "steps_per_s": 2.19, "commit_ce": 0.17939, "contact_bce": 0.23681, "goal_mse": 0.02418, "grad_norm": 3.93851, "jepa": 0.00035, "loss": 0.77348, "lr": 0.00012, "pr": 0.0782, "recon_mse": 0.05305, "sigreg": 0.16354, "state_mse": 0.00099, "step4_mse": 0.00181, "text_ce": 0.00014}
65
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3200, "goal_steps": 5600, "elapsed_s": 1570.17, "steps_per_s": 2.188, "commit_ce": 0.0999, "contact_bce": 0.24371, "goal_mse": 0.01803, "grad_norm": 6.95358, "jepa": 0.00042, "loss": 0.56136, "lr": 0.00011, "pr": 0.08593, "recon_mse": 0.07592, "sigreg": 0.15599, "state_mse": 0.00115, "step4_mse": 0.00215, "text_ce": 0.00016}
66
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3250, "goal_steps": 5600, "elapsed_s": 1593.02, "steps_per_s": 2.188, "commit_ce": 0.33175, "contact_bce": 0.20511, "goal_mse": 0.0086, "grad_norm": 4.87299, "jepa": 0.0003, "loss": 0.57953, "lr": 0.00011, "pr": 0.06546, "recon_mse": 0.05357, "sigreg": 0.17346, "state_mse": 0.00119, "step4_mse": 0.00325, "text_ce": 8e-05}
67
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3300, "goal_steps": 5600, "elapsed_s": 1615.9, "steps_per_s": 2.186, "commit_ce": 0.16044, "contact_bce": 0.28768, "goal_mse": 0.07359, "grad_norm": 28.5201, "jepa": 0.00036, "loss": 1.97645, "lr": 0.00011, "pr": 0.0821, "recon_mse": 0.04969, "sigreg": 0.15842, "state_mse": 0.0008, "step4_mse": 0.00218, "text_ce": 6e-05}
68
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3350, "goal_steps": 5600, "elapsed_s": 1638.76, "steps_per_s": 2.187, "commit_ce": 0.11147, "contact_bce": 0.24666, "goal_mse": 0.01833, "grad_norm": 5.51928, "jepa": 0.00029, "loss": 0.53557, "lr": 0.00011, "pr": 0.07, "recon_mse": 0.05428, "sigreg": 0.16773, "state_mse": 0.00123, "step4_mse": 0.0025, "text_ce": 0.00183}
69
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3400, "goal_steps": 5600, "elapsed_s": 1661.62, "steps_per_s": 2.187, "commit_ce": 0.06493, "contact_bce": 0.18626, "goal_mse": 0.02268, "grad_norm": 6.07611, "jepa": 0.00035, "loss": 0.49508, "lr": 0.0001, "pr": 0.0803, "recon_mse": 0.06263, "sigreg": 0.15875, "state_mse": 0.00121, "step4_mse": 0.0023, "text_ce": 3e-05}
70
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3450, "goal_steps": 5600, "elapsed_s": 1684.47, "steps_per_s": 2.188, "commit_ce": 0.14719, "contact_bce": 0.22921, "goal_mse": 0.01116, "grad_norm": 5.31518, "jepa": 0.00041, "loss": 0.75011, "lr": 0.0001, "pr": 0.10067, "recon_mse": 0.06309, "sigreg": 0.13596, "state_mse": 0.00127, "step4_mse": 0.00238, "text_ce": 0.00013}
71
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3500, "goal_steps": 5600, "elapsed_s": 1707.31, "steps_per_s": 2.189, "commit_ce": 0.20539, "contact_bce": 0.18535, "goal_mse": 0.016, "grad_norm": 6.16713, "jepa": 0.00037, "loss": 0.83709, "lr": 0.0001, "pr": 0.08629, "recon_mse": 0.05011, "sigreg": 0.149, "state_mse": 0.00083, "step4_mse": 0.00264, "text_ce": 0.00017}
72
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3550, "goal_steps": 5600, "elapsed_s": 1730.13, "steps_per_s": 2.191, "commit_ce": 0.23679, "contact_bce": 0.26021, "goal_mse": 0.00501, "grad_norm": 4.78782, "jepa": 0.00038, "loss": 1.08913, "lr": 9e-05, "pr": 0.09555, "recon_mse": 0.06479, "sigreg": 0.13158, "state_mse": 0.00073, "step4_mse": 0.00215, "text_ce": 4e-05}
73
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3600, "goal_steps": 5600, "elapsed_s": 1752.98, "steps_per_s": 2.188, "commit_ce": 0.08131, "contact_bce": 0.23462, "goal_mse": 0.01394, "grad_norm": 5.27649, "jepa": 0.00038, "loss": 0.48047, "lr": 9e-05, "pr": 0.09209, "recon_mse": 0.08099, "sigreg": 0.13509, "state_mse": 0.00127, "step4_mse": 0.00221, "text_ce": 8e-05}
74
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3650, "goal_steps": 5600, "elapsed_s": 1775.81, "steps_per_s": 2.19, "commit_ce": 0.18442, "contact_bce": 0.26237, "goal_mse": 0.01773, "grad_norm": 7.39635, "jepa": 0.00031, "loss": 1.13449, "lr": 9e-05, "pr": 0.07671, "recon_mse": 0.04142, "sigreg": 0.16509, "state_mse": 0.00062, "step4_mse": 0.00126, "text_ce": 3e-05}
75
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3700, "goal_steps": 5600, "elapsed_s": 1798.61, "steps_per_s": 2.194, "commit_ce": 0.19589, "contact_bce": 0.19063, "goal_mse": 0.0242, "grad_norm": 9.13761, "jepa": 0.00038, "loss": 0.69408, "lr": 8e-05, "pr": 0.09348, "recon_mse": 0.0554, "sigreg": 0.14013, "state_mse": 0.00087, "step4_mse": 0.0022, "text_ce": 2e-05}
76
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3750, "goal_steps": 5600, "elapsed_s": 1821.43, "steps_per_s": 2.191, "commit_ce": 0.23258, "contact_bce": 0.36022, "goal_mse": 0.02071, "grad_norm": 6.98649, "jepa": 0.00036, "loss": 0.86214, "lr": 8e-05, "pr": 0.09357, "recon_mse": 0.06177, "sigreg": 0.13243, "state_mse": 0.00103, "step4_mse": 0.0023, "text_ce": 3e-05}
77
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3800, "goal_steps": 5600, "elapsed_s": 1844.23, "steps_per_s": 2.193, "commit_ce": 0.23863, "contact_bce": 0.25294, "goal_mse": 0.02097, "grad_norm": 10.72185, "jepa": 0.00031, "loss": 0.5965, "lr": 8e-05, "pr": 0.07204, "recon_mse": 0.05719, "sigreg": 0.1615, "state_mse": 0.00111, "step4_mse": 0.00234, "text_ce": 4e-05}
78
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3850, "goal_steps": 5600, "elapsed_s": 1867.06, "steps_per_s": 2.19, "commit_ce": 0.21525, "contact_bce": 0.23725, "goal_mse": 0.02052, "grad_norm": 13.79049, "jepa": 0.00041, "loss": 1.6615, "lr": 8e-05, "pr": 0.09598, "recon_mse": 0.05938, "sigreg": 0.13346, "state_mse": 0.00146, "step4_mse": 0.00294, "text_ce": 3e-05}
79
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3900, "goal_steps": 5600, "elapsed_s": 1889.91, "steps_per_s": 2.188, "commit_ce": 0.27116, "contact_bce": 0.2288, "goal_mse": 0.02868, "grad_norm": 5.56726, "jepa": 0.0003, "loss": 0.62794, "lr": 7e-05, "pr": 0.07604, "recon_mse": 0.08933, "sigreg": 0.16655, "state_mse": 0.00091, "step4_mse": 0.00211, "text_ce": 4e-05}
80
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 3950, "goal_steps": 5600, "elapsed_s": 1912.76, "steps_per_s": 2.188, "commit_ce": 0.19693, "contact_bce": 0.22604, "goal_mse": 0.02609, "grad_norm": 13.32607, "jepa": 0.00034, "loss": 1.24157, "lr": 7e-05, "pr": 0.09059, "recon_mse": 0.03771, "sigreg": 0.14298, "state_mse": 0.00068, "step4_mse": 0.0021, "text_ce": 3e-05}
81
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4000, "goal_steps": 5600, "elapsed_s": 1935.62, "steps_per_s": 2.187, "commit_ce": 0.08585, "contact_bce": 0.25333, "goal_mse": 0.01871, "grad_norm": 6.92305, "jepa": 0.00039, "loss": 2.04987, "lr": 7e-05, "pr": 0.08967, "recon_mse": 0.06143, "sigreg": 0.13539, "state_mse": 0.00137, "step4_mse": 0.00293, "text_ce": 2e-05}
82
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4050, "goal_steps": 5600, "elapsed_s": 1958.47, "steps_per_s": 2.188, "commit_ce": 0.212, "contact_bce": 0.21367, "goal_mse": 0.00609, "grad_norm": 5.6922, "jepa": 0.00034, "loss": 0.45276, "lr": 7e-05, "pr": 0.07423, "recon_mse": 0.06863, "sigreg": 0.15894, "state_mse": 0.00083, "step4_mse": 0.00196, "text_ce": 2e-05}
83
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4100, "goal_steps": 5600, "elapsed_s": 1981.33, "steps_per_s": 2.188, "commit_ce": 0.26793, "contact_bce": 0.21339, "goal_mse": 0.03465, "grad_norm": 14.31906, "jepa": 0.00034, "loss": 0.58086, "lr": 6e-05, "pr": 0.08433, "recon_mse": 0.0671, "sigreg": 0.14691, "state_mse": 0.00059, "step4_mse": 0.00217, "text_ce": 2e-05}
84
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4150, "goal_steps": 5600, "elapsed_s": 2004.18, "steps_per_s": 2.188, "commit_ce": 0.16544, "contact_bce": 0.26901, "goal_mse": 0.01516, "grad_norm": 2.61141, "jepa": 0.00034, "loss": 0.53182, "lr": 6e-05, "pr": 0.08236, "recon_mse": 0.05361, "sigreg": 0.14574, "state_mse": 0.00114, "step4_mse": 0.00264, "text_ce": 5e-05}
85
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4200, "goal_steps": 5600, "elapsed_s": 2027.03, "steps_per_s": 2.188, "commit_ce": 0.1136, "contact_bce": 0.1789, "goal_mse": 0.02466, "grad_norm": 10.56715, "jepa": 0.00039, "loss": 0.76654, "lr": 6e-05, "pr": 0.09687, "recon_mse": 0.05019, "sigreg": 0.12746, "state_mse": 0.00121, "step4_mse": 0.00297, "text_ce": 2e-05}
86
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4250, "goal_steps": 5600, "elapsed_s": 2049.91, "steps_per_s": 2.186, "commit_ce": 0.24426, "contact_bce": 0.25715, "goal_mse": 0.00901, "grad_norm": 6.23569, "jepa": 0.00035, "loss": 0.62445, "lr": 6e-05, "pr": 0.0935, "recon_mse": 0.05397, "sigreg": 0.13989, "state_mse": 0.00053, "step4_mse": 0.0015, "text_ce": 2e-05}
87
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4300, "goal_steps": 5600, "elapsed_s": 2072.76, "steps_per_s": 2.188, "commit_ce": 0.18078, "contact_bce": 0.23848, "goal_mse": 0.01328, "grad_norm": 6.22292, "jepa": 0.00036, "loss": 0.43296, "lr": 5e-05, "pr": 0.10462, "recon_mse": 0.05615, "sigreg": 0.12081, "state_mse": 0.00044, "step4_mse": 0.00077, "text_ce": 1e-05}
88
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4350, "goal_steps": 5600, "elapsed_s": 2095.61, "steps_per_s": 2.188, "commit_ce": 0.10734, "contact_bce": 0.2184, "goal_mse": 0.00802, "grad_norm": 3.38702, "jepa": 0.0004, "loss": 0.50355, "lr": 5e-05, "pr": 0.10249, "recon_mse": 0.0733, "sigreg": 0.12138, "state_mse": 0.00083, "step4_mse": 0.00248, "text_ce": 2e-05}
89
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4400, "goal_steps": 5600, "elapsed_s": 2118.46, "steps_per_s": 2.188, "commit_ce": 0.17938, "contact_bce": 0.19201, "goal_mse": 0.00649, "grad_norm": 3.26897, "jepa": 0.00033, "loss": 0.56076, "lr": 5e-05, "pr": 0.08267, "recon_mse": 0.0388, "sigreg": 0.14926, "state_mse": 0.0011, "step4_mse": 0.00238, "text_ce": 2e-05}
90
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4450, "goal_steps": 5600, "elapsed_s": 2141.32, "steps_per_s": 2.188, "commit_ce": 0.01341, "contact_bce": 0.21278, "goal_mse": 0.00863, "grad_norm": 5.0816, "jepa": 0.00035, "loss": 0.41901, "lr": 5e-05, "pr": 0.08543, "recon_mse": 0.05805, "sigreg": 0.15449, "state_mse": 0.00126, "step4_mse": 0.00329, "text_ce": 2e-05}
91
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4500, "goal_steps": 5600, "elapsed_s": 2164.17, "steps_per_s": 2.188, "commit_ce": 0.14909, "contact_bce": 0.26111, "goal_mse": 0.01364, "grad_norm": 8.11348, "jepa": 0.00046, "loss": 0.69123, "lr": 5e-05, "pr": 0.11898, "recon_mse": 0.07608, "sigreg": 0.11378, "state_mse": 0.00099, "step4_mse": 0.00258, "text_ce": 1e-05}
92
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4550, "goal_steps": 5600, "elapsed_s": 2187.02, "steps_per_s": 2.188, "commit_ce": 0.16005, "contact_bce": 0.30168, "goal_mse": 0.02541, "grad_norm": 8.14659, "jepa": 0.00036, "loss": 0.73024, "lr": 4e-05, "pr": 0.08848, "recon_mse": 0.05601, "sigreg": 0.13492, "state_mse": 0.00104, "step4_mse": 0.00256, "text_ce": 2e-05}
93
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4600, "goal_steps": 5600, "elapsed_s": 2209.81, "steps_per_s": 2.194, "commit_ce": 0.09442, "contact_bce": 0.1717, "goal_mse": 0.01448, "grad_norm": 6.56101, "jepa": 0.00043, "loss": 0.60396, "lr": 4e-05, "pr": 0.10819, "recon_mse": 0.06558, "sigreg": 0.11786, "state_mse": 0.00163, "step4_mse": 0.00391, "text_ce": 1e-05}
94
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4650, "goal_steps": 5600, "elapsed_s": 2232.6, "steps_per_s": 2.194, "commit_ce": 0.15307, "contact_bce": 0.22979, "goal_mse": 0.0158, "grad_norm": 6.71283, "jepa": 0.00039, "loss": 0.59573, "lr": 4e-05, "pr": 0.10421, "recon_mse": 0.06187, "sigreg": 0.12001, "state_mse": 0.00094, "step4_mse": 0.00216, "text_ce": 1e-05}
95
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4700, "goal_steps": 5600, "elapsed_s": 2255.39, "steps_per_s": 2.194, "commit_ce": 0.24922, "contact_bce": 0.2316, "goal_mse": 0.03372, "grad_norm": 7.89936, "jepa": 0.00035, "loss": 0.57806, "lr": 4e-05, "pr": 0.09139, "recon_mse": 0.07394, "sigreg": 0.1277, "state_mse": 0.00079, "step4_mse": 0.00261, "text_ce": 2e-05}
96
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4750, "goal_steps": 5600, "elapsed_s": 2278.18, "steps_per_s": 2.194, "commit_ce": 0.14121, "contact_bce": 0.23942, "goal_mse": 0.0096, "grad_norm": 3.18005, "jepa": 0.00034, "loss": 0.45469, "lr": 4e-05, "pr": 0.09528, "recon_mse": 0.0462, "sigreg": 0.12757, "state_mse": 0.00079, "step4_mse": 0.00202, "text_ce": 1e-05}
97
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4800, "goal_steps": 5600, "elapsed_s": 2301.0, "steps_per_s": 2.191, "commit_ce": 0.18207, "contact_bce": 0.20716, "goal_mse": 0.00704, "grad_norm": 3.49327, "jepa": 0.00031, "loss": 0.43667, "lr": 4e-05, "pr": 0.08514, "recon_mse": 0.04594, "sigreg": 0.14425, "state_mse": 0.00081, "step4_mse": 0.00225, "text_ce": 1e-05}
98
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4850, "goal_steps": 5600, "elapsed_s": 2323.81, "steps_per_s": 2.192, "commit_ce": 0.1906, "contact_bce": 0.14996, "goal_mse": 0.02109, "grad_norm": 6.13158, "jepa": 0.00039, "loss": 0.54534, "lr": 3e-05, "pr": 0.09236, "recon_mse": 0.05358, "sigreg": 0.13399, "state_mse": 0.00085, "step4_mse": 0.00197, "text_ce": 1e-05}
99
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4900, "goal_steps": 5600, "elapsed_s": 2346.6, "steps_per_s": 2.194, "commit_ce": 0.13798, "contact_bce": 0.25962, "goal_mse": 0.00675, "grad_norm": 5.69474, "jepa": 0.00038, "loss": 0.55736, "lr": 3e-05, "pr": 0.08152, "recon_mse": 0.06202, "sigreg": 0.14366, "state_mse": 0.00111, "step4_mse": 0.0025, "text_ce": 2e-05}
100
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 4950, "goal_steps": 5600, "elapsed_s": 2369.41, "steps_per_s": 2.192, "commit_ce": 0.21104, "contact_bce": 0.16548, "goal_mse": 0.00732, "grad_norm": 4.41575, "jepa": 0.00033, "loss": 0.3917, "lr": 3e-05, "pr": 0.07957, "recon_mse": 0.03651, "sigreg": 0.15105, "state_mse": 0.00063, "step4_mse": 0.00135, "text_ce": 1e-05}
101
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5000, "goal_steps": 5600, "elapsed_s": 2392.24, "steps_per_s": 2.19, "commit_ce": 0.06463, "contact_bce": 0.22522, "goal_mse": 0.00733, "grad_norm": 3.52656, "jepa": 0.0004, "loss": 0.51159, "lr": 3e-05, "pr": 0.10355, "recon_mse": 0.07734, "sigreg": 0.1144, "state_mse": 0.00126, "step4_mse": 0.00312, "text_ce": 1e-05}
102
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5050, "goal_steps": 5600, "elapsed_s": 2415.06, "steps_per_s": 2.191, "commit_ce": 0.18707, "contact_bce": 0.24925, "goal_mse": 0.02252, "grad_norm": 5.10548, "jepa": 0.00038, "loss": 0.44398, "lr": 3e-05, "pr": 0.10334, "recon_mse": 0.07143, "sigreg": 0.12105, "state_mse": 0.00037, "step4_mse": 0.00098, "text_ce": 1e-05}
103
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5100, "goal_steps": 5600, "elapsed_s": 2437.88, "steps_per_s": 2.191, "commit_ce": 0.03139, "contact_bce": 0.21077, "goal_mse": 0.01243, "grad_norm": 3.81034, "jepa": 0.00037, "loss": 0.40174, "lr": 3e-05, "pr": 0.10147, "recon_mse": 0.04916, "sigreg": 0.12445, "state_mse": 0.00051, "step4_mse": 0.00156, "text_ce": 1e-05}
104
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5150, "goal_steps": 5600, "elapsed_s": 2460.67, "steps_per_s": 2.194, "commit_ce": 0.02529, "contact_bce": 0.20135, "goal_mse": 0.01372, "grad_norm": 9.27177, "jepa": 0.00048, "loss": 0.34637, "lr": 3e-05, "pr": 0.12393, "recon_mse": 0.062, "sigreg": 0.1057, "state_mse": 0.00044, "step4_mse": 0.00114, "text_ce": 1e-05}
105
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5200, "goal_steps": 5600, "elapsed_s": 2483.5, "steps_per_s": 2.191, "commit_ce": 0.23198, "contact_bce": 0.29216, "goal_mse": 0.01309, "grad_norm": 4.92782, "jepa": 0.0004, "loss": 0.75323, "lr": 3e-05, "pr": 0.10956, "recon_mse": 0.07416, "sigreg": 0.11102, "state_mse": 0.0009, "step4_mse": 0.00233, "text_ce": 1e-05}
106
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5250, "goal_steps": 5600, "elapsed_s": 2506.35, "steps_per_s": 2.188, "commit_ce": 0.08272, "contact_bce": 0.22104, "goal_mse": 0.01034, "grad_norm": 4.22359, "jepa": 0.00045, "loss": 0.40322, "lr": 3e-05, "pr": 0.12283, "recon_mse": 0.06035, "sigreg": 0.10338, "state_mse": 0.00062, "step4_mse": 0.00111, "text_ce": 1e-05}
107
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5300, "goal_steps": 5600, "elapsed_s": 2529.17, "steps_per_s": 2.191, "commit_ce": 0.05541, "contact_bce": 0.18195, "goal_mse": 0.02549, "grad_norm": 12.27848, "jepa": 0.0005, "loss": 0.42613, "lr": 3e-05, "pr": 0.11757, "recon_mse": 0.05013, "sigreg": 0.11457, "state_mse": 0.00059, "step4_mse": 0.00163, "text_ce": 2e-05}
108
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5350, "goal_steps": 5600, "elapsed_s": 2552.01, "steps_per_s": 2.189, "commit_ce": 0.20796, "contact_bce": 0.29698, "goal_mse": 0.01484, "grad_norm": 4.57219, "jepa": 0.00035, "loss": 0.58246, "lr": 3e-05, "pr": 0.08488, "recon_mse": 0.0525, "sigreg": 0.14428, "state_mse": 0.00059, "step4_mse": 0.00251, "text_ce": 1e-05}
109
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5400, "goal_steps": 5600, "elapsed_s": 2574.83, "steps_per_s": 2.191, "commit_ce": 0.01499, "contact_bce": 0.1879, "goal_mse": 0.00606, "grad_norm": 2.34234, "jepa": 0.00041, "loss": 0.35555, "lr": 2e-05, "pr": 0.11658, "recon_mse": 0.04158, "sigreg": 0.10903, "state_mse": 0.00056, "step4_mse": 0.00127, "text_ce": 4e-05}
110
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5450, "goal_steps": 5600, "elapsed_s": 2597.62, "steps_per_s": 2.194, "commit_ce": 0.01438, "contact_bce": 0.17689, "goal_mse": 0.01615, "grad_norm": 4.7073, "jepa": 0.00042, "loss": 0.44837, "lr": 2e-05, "pr": 0.11431, "recon_mse": 0.05133, "sigreg": 0.11096, "state_mse": 0.00091, "step4_mse": 0.00317, "text_ce": 1e-05}
111
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5500, "goal_steps": 5600, "elapsed_s": 2620.42, "steps_per_s": 2.193, "commit_ce": 0.02691, "contact_bce": 0.19507, "goal_mse": 0.00478, "grad_norm": 3.31971, "jepa": 0.00042, "loss": 0.4861, "lr": 2e-05, "pr": 0.09085, "recon_mse": 0.05136, "sigreg": 0.12879, "state_mse": 0.00151, "step4_mse": 0.00284, "text_ce": 0.0}
112
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5550, "goal_steps": 5600, "elapsed_s": 2643.28, "steps_per_s": 2.187, "commit_ce": 0.08373, "contact_bce": 0.20676, "goal_mse": 0.01886, "grad_norm": 7.65747, "jepa": 0.00045, "loss": 0.41116, "lr": 2e-05, "pr": 0.11432, "recon_mse": 0.06188, "sigreg": 0.10778, "state_mse": 0.00052, "step4_mse": 0.00169, "text_ce": 1e-05}
113
+ {"tag": "x1b-w5-slwm-1b-s2001", "step": 5600, "goal_steps": 5600, "elapsed_s": 2666.11, "steps_per_s": 2.19, "commit_ce": 0.1453, "contact_bce": 0.30349, "goal_mse": 0.00913, "grad_norm": 4.86116, "jepa": 0.00043, "loss": 0.57405, "lr": 2e-05, "pr": 0.11699, "recon_mse": 0.05393, "sigreg": 0.11152, "state_mse": 0.00071, "step4_mse": 0.00231, "text_ce": 0.0}
weights_bf16.npz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f56b803857a09e7e77e1bb501efdc8e47bcacb2230cf18a1ab44e1acca591f51
3
+ size 1602108632