| { |
| "package_scope": "historical_194m_run_registry", |
| "model_name": "SymbolicLight V1", |
| "intended_use": "Public registry for the main 194M experimental narrative absorbed from earlier DualPath-style release materials.", |
| "main_runs": [ |
| { |
| "run_name": "snn_auxce_s123", |
| "parameters_text": "194M", |
| "training_variant": "AuxCE", |
| "seed": 123, |
| "training_tokens_text": "3B bilingual tokens", |
| "val_ppl": 8.91, |
| "paper_reference": "Main 194M validation table" |
| }, |
| { |
| "run_name": "snn_auxce_s456", |
| "parameters_text": "194M", |
| "training_variant": "AuxCE", |
| "seed": 456, |
| "training_tokens_text": "3B bilingual tokens", |
| "val_ppl": 8.88, |
| "paper_reference": "Main 194M validation table" |
| }, |
| { |
| "run_name": "snn_noauxce_s42", |
| "parameters_text": "194M", |
| "training_variant": "noAuxCE", |
| "seed": 42, |
| "training_tokens_text": "3B bilingual tokens", |
| "val_ppl": 8.9, |
| "paper_reference": "Main 194M validation table" |
| }, |
| { |
| "run_name": "snn_noauxce_s123", |
| "parameters_text": "194M", |
| "training_variant": "noAuxCE", |
| "seed": 123, |
| "training_tokens_text": "3B bilingual tokens", |
| "val_ppl": 8.93, |
| "paper_reference": "Main 194M validation table" |
| } |
| ], |
| "dense_baselines": [ |
| { |
| "run_name": "gpt2_124m", |
| "parameters_text": "124M", |
| "training_variant": "dense baseline", |
| "seed": null, |
| "val_ppl": 8.96 |
| }, |
| { |
| "run_name": "gpt2_201m", |
| "parameters_text": "201M", |
| "training_variant": "dense baseline", |
| "seed": null, |
| "val_ppl": 8.27 |
| } |
| ], |
| "ablation_runs": [ |
| { |
| "run_name": "ablation_static_prior", |
| "parameters_text": "185M", |
| "training_variant": "0.5B token ablation", |
| "seed": 42, |
| "val_ppl": 21.26 |
| }, |
| { |
| "run_name": "ablation_no_attn", |
| "parameters_text": "173M", |
| "training_variant": "0.5B token ablation", |
| "seed": 42, |
| "val_ppl": 38.56 |
| }, |
| { |
| "run_name": "ablation_decay_only", |
| "parameters_text": "164M", |
| "training_variant": "0.5B token ablation", |
| "seed": 42, |
| "val_ppl": 25.27 |
| }, |
| { |
| "run_name": "ablation_topk_mask", |
| "parameters_text": "194M", |
| "training_variant": "0.5B token ablation", |
| "seed": 42, |
| "val_ppl": 43.88 |
| } |
| ], |
| "notes": [ |
| "This file preserves the historical 194M release narrative without recreating the old DualPath repository layout.", |
| "The current public package does not claim that all historical checkpoints listed here are publicly redistributed from this repository.", |
| "The current reproducibility boundary is still governed by REPRODUCIBILITY.md and the public artifacts actually shipped in this package." |
| ] |
| } |
|
|
