1.79 MB

Ctrl+K

1 contributor

History: 99 commits

PowerMachine

V6.5-attn: add upload report (post-upload artifact)

0cec8a1 verified 2 months ago

docs
Upload docs/analysis.md with huggingface_hub 3 months ago
model_final
V3.2: corrigir 6 bugs + meta_T_moved (T_tracking_loss) + re-teste com Xeon IPEX 2.8 + AMX 2 months ago
scripts
V6.5-attn cleanup: remove legacy scripts/upload_v6_5_7ds.py 2 months ago
src
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
.gitattributes

1.52 kB
initial commit 3 months ago
README.md

10 kB
Upload README.md with huggingface_hub 3 months ago
requirements.txt

113 Bytes
Upload requirements.txt with huggingface_hub 3 months ago
v6_4_report.json

7.32 kB
V6.4: refactored Kohonen SOM 4D (canonical user code, pgvector_lookup REMOVED) 2 months ago
v6_4_training_metrics.json

54.5 kB
V6.4: refactored Kohonen SOM 4D (canonical user code, pgvector_lookup REMOVED) 2 months ago
v6_4_upload_report.json

825 Bytes
V6.5: 1000 samples (500+500), MTP val+entropy, EWC+W8A8 eval investigation, integrate gru-ring modules 2 months ago
v6_5_7ds_attn_attention_eval.json

1.35 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_ewc_w8a8_benchmark.json

644 Bytes
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_module_analysis.json

4.62 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_reasoning_eval.json

2.21 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_report.json

11.9 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_script_activity.json

689 Bytes
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_training_metrics.json

260 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_upload_report.json

4.49 kB
V6.5-attn: add upload report (post-upload artifact) 2 months ago
v6_5_7ds_attn_verification.json

1.31 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago
v6_5_7ds_attn_w8a8_compression.json

23.3 kB
V6.5-attn: attention integrated to KLS (MultiHeadAttention n_heads=8), 100 samples/batch streaming, 5040 samples REAL (meta ≥5000), 864 neurons, SmoothQuant W8A8 (err=0.020), MTP K=6, tool_coordinator workers reactivated, 11/11 verification PASS, attention ACTIVE_AND_FUNCTIONAL, reasoning GOOD (100% answer/think), removed legacy v3/v4/v5/v6_1/v6_2/v6_3/v6_progressive + old v6_5_* + v6_5_7ds_* reports/scripts 2 months ago