snaykey commited on
Commit
25141df
·
verified ·
1 Parent(s): 260eed3

Update logbook: Repro: MemoryBench

Browse files
Files changed (2) hide show
  1. logbook.json +3 -3
  2. pages/protocol-and-artifacts/page.md +1027 -0
logbook.json CHANGED
@@ -8,7 +8,7 @@
8
  "icml2026-repro",
9
  "paper-If4X4W2HWx"
10
  ],
11
- "updated_at": "2026-07-17T12:22:29+00:00",
12
  "root": {
13
  "slug": "index",
14
  "title": "Repro: MemoryBench",
@@ -94,6 +94,6 @@
94
  }
95
  ]
96
  },
97
- "agent_view_tokens": 19367,
98
- "revision": "1784290949955517200"
99
  }
 
8
  "icml2026-repro",
9
  "paper-If4X4W2HWx"
10
  ],
11
+ "updated_at": "2026-07-17T12:23:11+00:00",
12
  "root": {
13
  "slug": "index",
14
  "title": "Repro: MemoryBench",
 
94
  }
95
  ]
96
  },
97
+ "agent_view_tokens": 19414,
98
+ "revision": "1784290991093832900"
99
  }
pages/protocol-and-artifacts/page.md CHANGED
@@ -1589,3 +1589,1030 @@ if __name__ == "__main__":
1589
  **📦 Artifact** `artifacts/memorybench-runtime/memorybench_reproduction_bundle.jsonl` · dataset · 5.5 MB
1590
 
1591
  trackio-local-path://artifacts/memorybench-runtime/memorybench_reproduction_bundle.jsonl
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1589
  **📦 Artifact** `artifacts/memorybench-runtime/memorybench_reproduction_bundle.jsonl` · dataset · 5.5 MB
1590
 
1591
  trackio-local-path://artifacts/memorybench-runtime/memorybench_reproduction_bundle.jsonl
1592
+
1593
+
1594
+ ---
1595
+ <!-- trackio-cell
1596
+ {"type": "figure", "id": "cell_196baff65708", "created_at": "2026-07-17T12:23:06+00:00", "title": "FINAL reproducibility manifest JSON"}
1597
+ -->
1598
+ ````html
1599
+ <div style=font-family:system-ui;padding:16px><b>Final non-circular manifest</b><p>31 hashed source and result files, 196 package versions, pricing, protocol, costs, commits, and deviations.</p></div>
1600
+ ````
1601
+
1602
+ ````raw
1603
+ {
1604
+ "generated_at_utc": "2026-07-17T12:22:01.752586+00:00",
1605
+ "paper": {
1606
+ "challenge_paper_number": 19023,
1607
+ "openreview_id": "If4X4W2HWx",
1608
+ "local_pdf": "paper.pdf",
1609
+ "numbering_note": "Current PDF uses Figure 3 for Claim 4, Figure 4/Table 19 for Claim 5, and Tables 16-17 for Claim 6; challenge wording cites Figures 2-3 and Table 11 from an earlier revision."
1610
+ },
1611
+ "source": {
1612
+ "workspace_commit_at_manifest": "4717f93eafd34b0b12339024f8d8a20be005a551",
1613
+ "upstream_repository": "THUIR/MemoryBench",
1614
+ "upstream_commit": "5eafebca4e9ffbb2f0087ade13c498cf95fbc09a",
1615
+ "upstream_status": "",
1616
+ "dataset_repository": "THUIR/MemoryBench",
1617
+ "dataset_revision_observed_in_cache": "3acd60a4bd35b43b408f0e6db4c5f1e88df5e96d"
1618
+ },
1619
+ "runtime": {
1620
+ "platform": "Windows-11-10.0.26200-SP0",
1621
+ "python": "3.12.13 (main, Mar 3 2026, 15:01:35) [MSC v.1944 64 bit (AMD64)]",
1622
+ "executable": "C:\\Users\\User\\Documents\\CC Local\\icml-repro\\repro-memorybench\\MemoryBench\\.venv\\Scripts\\python.exe",
1623
+ "packages": [
1624
+ {
1625
+ "name": "absl-py",
1626
+ "version": "2.5.0"
1627
+ },
1628
+ {
1629
+ "name": "accelerate",
1630
+ "version": "1.14.0"
1631
+ },
1632
+ {
1633
+ "name": "aiohappyeyeballs",
1634
+ "version": "2.7.1"
1635
+ },
1636
+ {
1637
+ "name": "aiohttp",
1638
+ "version": "3.14.1"
1639
+ },
1640
+ {
1641
+ "name": "aiosignal",
1642
+ "version": "1.4.0"
1643
+ },
1644
+ {
1645
+ "name": "aiosqlite",
1646
+ "version": "0.22.1"
1647
+ },
1648
+ {
1649
+ "name": "altair",
1650
+ "version": "6.2.2"
1651
+ },
1652
+ {
1653
+ "name": "annotated-doc",
1654
+ "version": "0.0.4"
1655
+ },
1656
+ {
1657
+ "name": "annotated-types",
1658
+ "version": "0.7.0"
1659
+ },
1660
+ {
1661
+ "name": "anthropic",
1662
+ "version": "0.117.0"
1663
+ },
1664
+ {
1665
+ "name": "anyio",
1666
+ "version": "4.14.2"
1667
+ },
1668
+ {
1669
+ "name": "attrs",
1670
+ "version": "26.1.0"
1671
+ },
1672
+ {
1673
+ "name": "backoff",
1674
+ "version": "2.2.1"
1675
+ },
1676
+ {
1677
+ "name": "banks",
1678
+ "version": "2.4.5"
1679
+ },
1680
+ {
1681
+ "name": "bcrypt",
1682
+ "version": "5.0.0"
1683
+ },
1684
+ {
1685
+ "name": "bert-score",
1686
+ "version": "0.3.13"
1687
+ },
1688
+ {
1689
+ "name": "blinker",
1690
+ "version": "1.9.0"
1691
+ },
1692
+ {
1693
+ "name": "build",
1694
+ "version": "1.5.0"
1695
+ },
1696
+ {
1697
+ "name": "cachetools",
1698
+ "version": "7.1.4"
1699
+ },
1700
+ {
1701
+ "name": "certifi",
1702
+ "version": "2026.6.17"
1703
+ },
1704
+ {
1705
+ "name": "charset-normalizer",
1706
+ "version": "3.4.9"
1707
+ },
1708
+ {
1709
+ "name": "chinese2digits",
1710
+ "version": "6.0.2"
1711
+ },
1712
+ {
1713
+ "name": "chromadb",
1714
+ "version": "1.5.9"
1715
+ },
1716
+ {
1717
+ "name": "click",
1718
+ "version": "8.4.2"
1719
+ },
1720
+ {
1721
+ "name": "cn2an",
1722
+ "version": "0.5.24"
1723
+ },
1724
+ {
1725
+ "name": "colorama",
1726
+ "version": "0.4.6"
1727
+ },
1728
+ {
1729
+ "name": "contourpy",
1730
+ "version": "1.3.3"
1731
+ },
1732
+ {
1733
+ "name": "cycler",
1734
+ "version": "0.12.1"
1735
+ },
1736
+ {
1737
+ "name": "dataclasses-json",
1738
+ "version": "0.6.7"
1739
+ },
1740
+ {
1741
+ "name": "datasets",
1742
+ "version": "5.0.0"
1743
+ },
1744
+ {
1745
+ "name": "defusedxml",
1746
+ "version": "0.7.1"
1747
+ },
1748
+ {
1749
+ "name": "Deprecated",
1750
+ "version": "1.3.1"
1751
+ },
1752
+ {
1753
+ "name": "dill",
1754
+ "version": "0.4.1"
1755
+ },
1756
+ {
1757
+ "name": "dirtyjson",
1758
+ "version": "1.0.8"
1759
+ },
1760
+ {
1761
+ "name": "distro",
1762
+ "version": "1.9.0"
1763
+ },
1764
+ {
1765
+ "name": "docstring_parser",
1766
+ "version": "0.18.0"
1767
+ },
1768
+ {
1769
+ "name": "durationpy",
1770
+ "version": "0.10"
1771
+ },
1772
+ {
1773
+ "name": "evaluate",
1774
+ "version": "0.4.6"
1775
+ },
1776
+ {
1777
+ "name": "faiss-cpu",
1778
+ "version": "1.14.3"
1779
+ },
1780
+ {
1781
+ "name": "fastuuid",
1782
+ "version": "0.14.0"
1783
+ },
1784
+ {
1785
+ "name": "filelock",
1786
+ "version": "3.30.2"
1787
+ },
1788
+ {
1789
+ "name": "filetype",
1790
+ "version": "1.2.0"
1791
+ },
1792
+ {
1793
+ "name": "FlagEmbedding",
1794
+ "version": "1.4.0"
1795
+ },
1796
+ {
1797
+ "name": "flatbuffers",
1798
+ "version": "25.12.19"
1799
+ },
1800
+ {
1801
+ "name": "fonttools",
1802
+ "version": "4.63.0"
1803
+ },
1804
+ {
1805
+ "name": "frozenlist",
1806
+ "version": "1.8.0"
1807
+ },
1808
+ {
1809
+ "name": "fsspec",
1810
+ "version": "2026.4.0"
1811
+ },
1812
+ {
1813
+ "name": "gitdb",
1814
+ "version": "4.0.12"
1815
+ },
1816
+ {
1817
+ "name": "GitPython",
1818
+ "version": "3.1.52"
1819
+ },
1820
+ {
1821
+ "name": "googleapis-common-protos",
1822
+ "version": "1.75.0"
1823
+ },
1824
+ {
1825
+ "name": "greenlet",
1826
+ "version": "3.5.3"
1827
+ },
1828
+ {
1829
+ "name": "griffe",
1830
+ "version": "2.1.0"
1831
+ },
1832
+ {
1833
+ "name": "griffecli",
1834
+ "version": "2.1.0"
1835
+ },
1836
+ {
1837
+ "name": "griffelib",
1838
+ "version": "2.1.0"
1839
+ },
1840
+ {
1841
+ "name": "grpcio",
1842
+ "version": "1.82.1"
1843
+ },
1844
+ {
1845
+ "name": "h11",
1846
+ "version": "0.16.0"
1847
+ },
1848
+ {
1849
+ "name": "h2",
1850
+ "version": "4.3.0"
1851
+ },
1852
+ {
1853
+ "name": "hf-xet",
1854
+ "version": "1.5.2"
1855
+ },
1856
+ {
1857
+ "name": "hpack",
1858
+ "version": "4.2.0"
1859
+ },
1860
+ {
1861
+ "name": "httpcore",
1862
+ "version": "1.0.9"
1863
+ },
1864
+ {
1865
+ "name": "httptools",
1866
+ "version": "0.8.0"
1867
+ },
1868
+ {
1869
+ "name": "httpx",
1870
+ "version": "0.28.1"
1871
+ },
1872
+ {
1873
+ "name": "huggingface_hub",
1874
+ "version": "1.24.0"
1875
+ },
1876
+ {
1877
+ "name": "hyperframe",
1878
+ "version": "6.1.0"
1879
+ },
1880
+ {
1881
+ "name": "idna",
1882
+ "version": "3.18"
1883
+ },
1884
+ {
1885
+ "name": "importlib_metadata",
1886
+ "version": "8.9.0"
1887
+ },
1888
+ {
1889
+ "name": "importlib_resources",
1890
+ "version": "7.1.0"
1891
+ },
1892
+ {
1893
+ "name": "ir_datasets",
1894
+ "version": "0.6.2"
1895
+ },
1896
+ {
1897
+ "name": "itsdangerous",
1898
+ "version": "2.2.0"
1899
+ },
1900
+ {
1901
+ "name": "jieba",
1902
+ "version": "0.42.1"
1903
+ },
1904
+ {
1905
+ "name": "Jinja2",
1906
+ "version": "3.1.6"
1907
+ },
1908
+ {
1909
+ "name": "jiter",
1910
+ "version": "0.16.0"
1911
+ },
1912
+ {
1913
+ "name": "joblib",
1914
+ "version": "1.5.3"
1915
+ },
1916
+ {
1917
+ "name": "jsonlines",
1918
+ "version": "4.0.0"
1919
+ },
1920
+ {
1921
+ "name": "jsonschema",
1922
+ "version": "4.26.0"
1923
+ },
1924
+ {
1925
+ "name": "jsonschema-specifications",
1926
+ "version": "2025.9.1"
1927
+ },
1928
+ {
1929
+ "name": "kiwisolver",
1930
+ "version": "1.5.0"
1931
+ },
1932
+ {
1933
+ "name": "kubernetes",
1934
+ "version": "36.0.3"
1935
+ },
1936
+ {
1937
+ "name": "Levenshtein",
1938
+ "version": "0.27.3"
1939
+ },
1940
+ {
1941
+ "name": "litellm",
1942
+ "version": "1.92.0"
1943
+ },
1944
+ {
1945
+ "name": "llama-index",
1946
+ "version": "0.14.23"
1947
+ },
1948
+ {
1949
+ "name": "llama-index-core",
1950
+ "version": "0.14.23"
1951
+ },
1952
+ {
1953
+ "name": "llama-index-embeddings-openai",
1954
+ "version": "0.6.0"
1955
+ },
1956
+ {
1957
+ "name": "llama-index-instrumentation",
1958
+ "version": "0.5.0"
1959
+ },
1960
+ {
1961
+ "name": "llama-index-llms-openai",
1962
+ "version": "0.7.9"
1963
+ },
1964
+ {
1965
+ "name": "llama-index-workflows",
1966
+ "version": "2.22.2"
1967
+ },
1968
+ {
1969
+ "name": "lxml",
1970
+ "version": "6.1.1"
1971
+ },
1972
+ {
1973
+ "name": "lz4",
1974
+ "version": "4.4.5"
1975
+ },
1976
+ {
1977
+ "name": "markdown-it-py",
1978
+ "version": "4.2.0"
1979
+ },
1980
+ {
1981
+ "name": "MarkupSafe",
1982
+ "version": "3.0.3"
1983
+ },
1984
+ {
1985
+ "name": "marshmallow",
1986
+ "version": "3.26.2"
1987
+ },
1988
+ {
1989
+ "name": "matplotlib",
1990
+ "version": "3.11.0"
1991
+ },
1992
+ {
1993
+ "name": "mdurl",
1994
+ "version": "0.1.2"
1995
+ },
1996
+ {
1997
+ "name": "mem0ai",
1998
+ "version": "0.1.116"
1999
+ },
2000
+ {
2001
+ "name": "mmh3",
2002
+ "version": "5.2.1"
2003
+ },
2004
+ {
2005
+ "name": "mpmath",
2006
+ "version": "1.3.0"
2007
+ },
2008
+ {
2009
+ "name": "multidict",
2010
+ "version": "6.7.1"
2011
+ },
2012
+ {
2013
+ "name": "multiprocess",
2014
+ "version": "0.70.19"
2015
+ },
2016
+ {
2017
+ "name": "mypy_extensions",
2018
+ "version": "1.1.0"
2019
+ },
2020
+ {
2021
+ "name": "narwhals",
2022
+ "version": "2.24.0"
2023
+ },
2024
+ {
2025
+ "name": "nest-asyncio",
2026
+ "version": "1.6.0"
2027
+ },
2028
+ {
2029
+ "name": "networkx",
2030
+ "version": "3.6.1"
2031
+ },
2032
+ {
2033
+ "name": "nltk",
2034
+ "version": "3.10.0"
2035
+ },
2036
+ {
2037
+ "name": "numpy",
2038
+ "version": "2.5.1"
2039
+ },
2040
+ {
2041
+ "name": "oauthlib",
2042
+ "version": "3.3.1"
2043
+ },
2044
+ {
2045
+ "name": "onnxruntime",
2046
+ "version": "1.27.0"
2047
+ },
2048
+ {
2049
+ "name": "openai",
2050
+ "version": "1.99.9"
2051
+ },
2052
+ {
2053
+ "name": "opentelemetry-api",
2054
+ "version": "1.44.0"
2055
+ },
2056
+ {
2057
+ "name": "opentelemetry-exporter-otlp-proto-common",
2058
+ "version": "1.44.0"
2059
+ },
2060
+ {
2061
+ "name": "opentelemetry-exporter-otlp-proto-grpc",
2062
+ "version": "1.44.0"
2063
+ },
2064
+ {
2065
+ "name": "opentelemetry-proto",
2066
+ "version": "1.44.0"
2067
+ },
2068
+ {
2069
+ "name": "opentelemetry-sdk",
2070
+ "version": "1.44.0"
2071
+ },
2072
+ {
2073
+ "name": "opentelemetry-semantic-conventions",
2074
+ "version": "0.65b0"
2075
+ },
2076
+ {
2077
+ "name": "orjson",
2078
+ "version": "3.11.9"
2079
+ },
2080
+ {
2081
+ "name": "overrides",
2082
+ "version": "7.7.0"
2083
+ },
2084
+ {
2085
+ "name": "packaging",
2086
+ "version": "26.2"
2087
+ },
2088
+ {
2089
+ "name": "pandas",
2090
+ "version": "3.0.3"
2091
+ },
2092
+ {
2093
+ "name": "peft",
2094
+ "version": "0.19.1"
2095
+ },
2096
+ {
2097
+ "name": "pillow",
2098
+ "version": "12.3.0"
2099
+ },
2100
+ {
2101
+ "name": "pip",
2102
+ "version": "25.0.1"
2103
+ },
2104
+ {
2105
+ "name": "platformdirs",
2106
+ "version": "4.10.0"
2107
+ },
2108
+ {
2109
+ "name": "portalocker",
2110
+ "version": "3.2.0"
2111
+ },
2112
+ {
2113
+ "name": "posthog",
2114
+ "version": "7.25.0"
2115
+ },
2116
+ {
2117
+ "name": "proces",
2118
+ "version": "0.1.7"
2119
+ },
2120
+ {
2121
+ "name": "propcache",
2122
+ "version": "0.5.2"
2123
+ },
2124
+ {
2125
+ "name": "protobuf",
2126
+ "version": "5.29.6"
2127
+ },
2128
+ {
2129
+ "name": "psutil",
2130
+ "version": "7.2.2"
2131
+ },
2132
+ {
2133
+ "name": "pyarrow",
2134
+ "version": "24.0.0"
2135
+ },
2136
+ {
2137
+ "name": "pybase64",
2138
+ "version": "1.4.3"
2139
+ },
2140
+ {
2141
+ "name": "pydantic",
2142
+ "version": "2.13.4"
2143
+ },
2144
+ {
2145
+ "name": "pydantic-settings",
2146
+ "version": "2.14.2"
2147
+ },
2148
+ {
2149
+ "name": "pydantic_core",
2150
+ "version": "2.46.4"
2151
+ },
2152
+ {
2153
+ "name": "pydeck",
2154
+ "version": "0.9.3"
2155
+ },
2156
+ {
2157
+ "name": "Pygments",
2158
+ "version": "2.20.0"
2159
+ },
2160
+ {
2161
+ "name": "pyparsing",
2162
+ "version": "3.3.2"
2163
+ },
2164
+ {
2165
+ "name": "pyphen",
2166
+ "version": "0.17.2"
2167
+ },
2168
+ {
2169
+ "name": "PyPika",
2170
+ "version": "0.51.1"
2171
+ },
2172
+ {
2173
+ "name": "pyproject_hooks",
2174
+ "version": "1.2.0"
2175
+ },
2176
+ {
2177
+ "name": "python-dateutil",
2178
+ "version": "2.9.0.post0"
2179
+ },
2180
+ {
2181
+ "name": "python-dotenv",
2182
+ "version": "1.2.2"
2183
+ },
2184
+ {
2185
+ "name": "python-multipart",
2186
+ "version": "0.0.32"
2187
+ },
2188
+ {
2189
+ "name": "pytz",
2190
+ "version": "2026.2"
2191
+ },
2192
+ {
2193
+ "name": "pywin32",
2194
+ "version": "312"
2195
+ },
2196
+ {
2197
+ "name": "PyYAML",
2198
+ "version": "6.0.3"
2199
+ },
2200
+ {
2201
+ "name": "qdrant-client",
2202
+ "version": "1.18.0"
2203
+ },
2204
+ {
2205
+ "name": "rank-bm25",
2206
+ "version": "0.2.2"
2207
+ },
2208
+ {
2209
+ "name": "RapidFuzz",
2210
+ "version": "3.14.5"
2211
+ },
2212
+ {
2213
+ "name": "referencing",
2214
+ "version": "0.37.0"
2215
+ },
2216
+ {
2217
+ "name": "regex",
2218
+ "version": "2026.7.10"
2219
+ },
2220
+ {
2221
+ "name": "requests",
2222
+ "version": "2.34.2"
2223
+ },
2224
+ {
2225
+ "name": "requests-oauthlib",
2226
+ "version": "2.0.0"
2227
+ },
2228
+ {
2229
+ "name": "rich",
2230
+ "version": "15.0.0"
2231
+ },
2232
+ {
2233
+ "name": "rouge",
2234
+ "version": "1.0.1"
2235
+ },
2236
+ {
2237
+ "name": "rouge_score",
2238
+ "version": "0.1.2"
2239
+ },
2240
+ {
2241
+ "name": "rpds-py",
2242
+ "version": "2026.6.3"
2243
+ },
2244
+ {
2245
+ "name": "safetensors",
2246
+ "version": "0.8.0"
2247
+ },
2248
+ {
2249
+ "name": "scikit-learn",
2250
+ "version": "1.9.0"
2251
+ },
2252
+ {
2253
+ "name": "scipy",
2254
+ "version": "1.18.0"
2255
+ },
2256
+ {
2257
+ "name": "sentence-transformers",
2258
+ "version": "5.6.0"
2259
+ },
2260
+ {
2261
+ "name": "sentencepiece",
2262
+ "version": "0.2.2"
2263
+ },
2264
+ {
2265
+ "name": "setuptools",
2266
+ "version": "83.0.0"
2267
+ },
2268
+ {
2269
+ "name": "shellingham",
2270
+ "version": "1.5.4"
2271
+ },
2272
+ {
2273
+ "name": "six",
2274
+ "version": "1.17.0"
2275
+ },
2276
+ {
2277
+ "name": "smmap",
2278
+ "version": "5.0.3"
2279
+ },
2280
+ {
2281
+ "name": "sniffio",
2282
+ "version": "1.3.1"
2283
+ },
2284
+ {
2285
+ "name": "SQLAlchemy",
2286
+ "version": "2.0.51"
2287
+ },
2288
+ {
2289
+ "name": "starlette",
2290
+ "version": "1.3.1"
2291
+ },
2292
+ {
2293
+ "name": "streamlit",
2294
+ "version": "1.59.2"
2295
+ },
2296
+ {
2297
+ "name": "sympy",
2298
+ "version": "1.14.0"
2299
+ },
2300
+ {
2301
+ "name": "tenacity",
2302
+ "version": "9.1.4"
2303
+ },
2304
+ {
2305
+ "name": "termcolor",
2306
+ "version": "3.3.0"
2307
+ },
2308
+ {
2309
+ "name": "textstat",
2310
+ "version": "0.7.13"
2311
+ },
2312
+ {
2313
+ "name": "threadpoolctl",
2314
+ "version": "3.6.0"
2315
+ },
2316
+ {
2317
+ "name": "tiktoken",
2318
+ "version": "0.13.0"
2319
+ },
2320
+ {
2321
+ "name": "tinytag",
2322
+ "version": "2.2.1"
2323
+ },
2324
+ {
2325
+ "name": "tokenizers",
2326
+ "version": "0.22.2"
2327
+ },
2328
+ {
2329
+ "name": "toml",
2330
+ "version": "0.10.2"
2331
+ },
2332
+ {
2333
+ "name": "torch",
2334
+ "version": "2.13.0"
2335
+ },
2336
+ {
2337
+ "name": "tqdm",
2338
+ "version": "4.68.4"
2339
+ },
2340
+ {
2341
+ "name": "transformers",
2342
+ "version": "5.14.1"
2343
+ },
2344
+ {
2345
+ "name": "typer",
2346
+ "version": "0.27.0"
2347
+ },
2348
+ {
2349
+ "name": "typing-inspect",
2350
+ "version": "0.9.0"
2351
+ },
2352
+ {
2353
+ "name": "typing-inspection",
2354
+ "version": "0.4.2"
2355
+ },
2356
+ {
2357
+ "name": "typing_extensions",
2358
+ "version": "4.16.0"
2359
+ },
2360
+ {
2361
+ "name": "tzdata",
2362
+ "version": "2026.3"
2363
+ },
2364
+ {
2365
+ "name": "urllib3",
2366
+ "version": "2.7.0"
2367
+ },
2368
+ {
2369
+ "name": "uvicorn",
2370
+ "version": "0.51.0"
2371
+ },
2372
+ {
2373
+ "name": "watchdog",
2374
+ "version": "6.0.0"
2375
+ },
2376
+ {
2377
+ "name": "watchfiles",
2378
+ "version": "1.2.0"
2379
+ },
2380
+ {
2381
+ "name": "websocket-client",
2382
+ "version": "1.9.0"
2383
+ },
2384
+ {
2385
+ "name": "websockets",
2386
+ "version": "16.1"
2387
+ },
2388
+ {
2389
+ "name": "Whoosh",
2390
+ "version": "2.7.4"
2391
+ },
2392
+ {
2393
+ "name": "wrapt",
2394
+ "version": "2.2.2"
2395
+ },
2396
+ {
2397
+ "name": "xxhash",
2398
+ "version": "3.8.1"
2399
+ },
2400
+ {
2401
+ "name": "yarl",
2402
+ "version": "1.24.2"
2403
+ },
2404
+ {
2405
+ "name": "zipp",
2406
+ "version": "4.1.0"
2407
+ }
2408
+ ]
2409
+ },
2410
+ "inference": {
2411
+ "model": "Qwen/Qwen3-8B",
2412
+ "routed_model": "Qwen/Qwen3-8B:nscale",
2413
+ "provider": "nscale via HF Inference Providers",
2414
+ "live_price_checked_before_each proxy start": {
2415
+ "input_usd_per_million_tokens": 0.07,
2416
+ "output_usd_per_million_tokens": 0.18,
2417
+ "context_length": 40960
2418
+ },
2419
+ "temperature": 0.1,
2420
+ "top_p": 0.1,
2421
+ "seed": 42,
2422
+ "retrieve_k": 5,
2423
+ "qwen_mode": "non-thinking, using /no_think provider prompt switch",
2424
+ "embeddings": {
2425
+ "A-Mem_and_MemoryOS": "all-MiniLM-L6-v2 local",
2426
+ "Mem0": "Qwen/Qwen3-Embedding-0.6B local, 1024 dimensions"
2427
+ }
2428
+ },
2429
+ "cost_controls": {
2430
+ "call_cap": 500,
2431
+ "automatic_review_checkpoint_usd": 0.5,
2432
+ "absolute_hard_cap_usd": 2.0,
2433
+ "provider_usage_required": true,
2434
+ "ledger": {
2435
+ "actual_cost_usd": 0.05059508,
2436
+ "calls": 268,
2437
+ "last_updated_utc": "2026-07-17T11:57:49.284606+00:00"
2438
+ },
2439
+ "completed_result_cost_usd": 0.04947408,
2440
+ "difference_includes_canaries_failed_or_orphaned_attempts_usd": 0.001121000000000004,
2441
+ "orphaned_but_fully_ledgered_calls": 7,
2442
+ "orphan_reason": "Trackio Windows CP-1252 reader crashed on UTF-8 MemoryOS output after provider calls; calls remained in calls.jsonl and the run was repeated only after capture encoding was made compatible."
2443
+ },
2444
+ "protocol": {
2445
+ "claims_4_5": "scaled off-policy; 5 seed-42 train and 5 seed-42 test cases on LexEval-QA and Locomo-0; identical slice for five baselines",
2446
+ "locomo_corpus": "all evidence turns for selected test questions plus seed-42 distractors, 40 turns total",
2447
+ "claim_6": "Appendix A.3.5 paired first response vs re-answer after full released dialogue on 5 LexEval-QA training cases",
2448
+ "claim_3": "runtime schema probe on complete released LexEval-QA and Locomo-0 training splits plus bounded Table 1 comparison"
2449
+ },
2450
+ "declared_deviations_and_limits": [
2451
+ "Scaled 5/5 samples and one random seed, not the paper's full datasets; no confidence intervals or significance test are claimed.",
2452
+ "Locomo uses an evidence-preserving 40-turn slice rather than all 419 corpus turns so all five systems fit the 500-call safety cap.",
2453
+ "Qwen3-8B is served per call by HF nscale rather than local vLLM on an A800 GPU.",
2454
+ "The provider lacks a chat-template flag, so Qwen3 non-thinking mode is selected with its /no_think prompt switch.",
2455
+ "The provider reports structured_output=false; only the transport response_format field is removed while MemoryBench's JSON schema instructions remain in prompts.",
2456
+ "Trackio console capture uses CP-1252 with replacement on Windows; persisted JSON artifacts remain UTF-8.",
2457
+ "Claim 3's comparative negative is limited to the seven prior benchmarks named in paper Table 1.",
2458
+ "Observed wall time includes local hardware, network, provider latency, and cold-cache effects; it is directly comparable within this campaign but not numerically identical to the paper's hardware."
2459
+ ],
2460
+ "hashes": [
2461
+ {
2462
+ "path": "artifacts\\memorybench-runtime\\analysis\\claim3_runtime.json",
2463
+ "bytes": 7384,
2464
+ "sha256": "f1b6706ebb1dbe7a6287c9459c4f6210d01f8b65fac82bbcfa7359ce5e4534a2"
2465
+ },
2466
+ {
2467
+ "path": "artifacts\\memorybench-runtime\\analysis\\claim6_paired.csv",
2468
+ "bytes": 357,
2469
+ "sha256": "1ada5a8e89be4bc20d3660c5ba6a9ed50257ea6fa18b613ef632c05cf6fefdd1"
2470
+ },
2471
+ {
2472
+ "path": "artifacts\\memorybench-runtime\\analysis\\completed_runs.csv",
2473
+ "bytes": 2412,
2474
+ "sha256": "977966a438a4805195ea4280f4ec78c51e0542871288a13056de9c7d93f251bc"
2475
+ },
2476
+ {
2477
+ "path": "artifacts\\memorybench-runtime\\analysis\\summary.json",
2478
+ "bytes": 2750,
2479
+ "sha256": "14f91fd115ac9fed7ef407dd21273b73f696f277d1c47b4e5934712d07ebc7ec"
2480
+ },
2481
+ {
2482
+ "path": "artifacts\\memorybench-runtime\\calls.jsonl",
2483
+ "bytes": 113392,
2484
+ "sha256": "4f7355792a6b77d701c634202aa839bef973170c65396bded35a9f6f473f463c"
2485
+ },
2486
+ {
2487
+ "path": "artifacts\\memorybench-runtime\\cost_ledger.json",
2488
+ "bytes": 113,
2489
+ "sha256": "0dad06f6d94f48b531be674ca05f9b15ef4d6be7599d7400692d2d4bb19f40da"
2490
+ },
2491
+ {
2492
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112122Z\\results.json",
2493
+ "bytes": 3477,
2494
+ "sha256": "5b939b1ae805713ff8f6903f00691a409999469ed1579d4a58e98ca89d0293c2"
2495
+ },
2496
+ {
2497
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112215Z\\results.json",
2498
+ "bytes": 59470,
2499
+ "sha256": "e9bd60e583df9e30504a46a0d85c7ed3f8a774272748be6095ac05229e95fa2c"
2500
+ },
2501
+ {
2502
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112335Z\\results.json",
2503
+ "bytes": 510432,
2504
+ "sha256": "7b6afa7df9b099d0f51e7877c46a674d746e359aa87bb8423879eaceb1d9238f"
2505
+ },
2506
+ {
2507
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112524Z\\results.json",
2508
+ "bytes": 517173,
2509
+ "sha256": "5067cb6adc218370be8d9e975f4928aae51f8c9e71c9d1b9803ea439efc2746d"
2510
+ },
2511
+ {
2512
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112623Z\\results.json",
2513
+ "bytes": 124196,
2514
+ "sha256": "8e4ba802dd0af27ceabba86369f53477bac0823f47cd1ba60240a1031acd359f"
2515
+ },
2516
+ {
2517
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112816Z\\results.json",
2518
+ "bytes": 20082,
2519
+ "sha256": "bc42fbb50e45db375539fce1a69ba1b2545fe46c57ad228ad6ccd8871f38d7d2"
2520
+ },
2521
+ {
2522
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T112951Z\\results.json",
2523
+ "bytes": 3735,
2524
+ "sha256": "c836088be40c611702bf23529365060c0a09048e95c32d5338a1615e37f59baa"
2525
+ },
2526
+ {
2527
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T113141Z\\results.json",
2528
+ "bytes": 8762,
2529
+ "sha256": "5758375d0b35e9a4a4e6658da222779af8932209c82ab1b2b727d82173d9bb43"
2530
+ },
2531
+ {
2532
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T113256Z\\results.json",
2533
+ "bytes": 474593,
2534
+ "sha256": "1a31bb0fb303a5e616af50651d16c8885d3eb33b704edea539dea07131576b9a"
2535
+ },
2536
+ {
2537
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T113603Z\\results.json",
2538
+ "bytes": 71743,
2539
+ "sha256": "ba7195c184154a52d5e35a8ab1bc28f2393d4525749921b47b81478a3a2428b7"
2540
+ },
2541
+ {
2542
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T113848Z\\results.json",
2543
+ "bytes": 30215,
2544
+ "sha256": "750d506baa359903b88b6d79386c82e974c19acdca82e7640d7b6edadbad8e58"
2545
+ },
2546
+ {
2547
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T114619Z\\results.json",
2548
+ "bytes": 12628,
2549
+ "sha256": "a2082062bab7d54166d0e844e0ab185757d17185ddce4915b11f7e5d23da8437"
2550
+ },
2551
+ {
2552
+ "path": "artifacts\\memorybench-runtime\\runs\\20260717T115724Z\\feedback_effect.json",
2553
+ "bytes": 240954,
2554
+ "sha256": "2faa64d2869c0e519d488beb1f5bb0add815d11c32464a800a18a2abe22f04bb"
2555
+ },
2556
+ {
2557
+ "path": "experiments\\analyze_runtime.py",
2558
+ "bytes": 9124,
2559
+ "sha256": "365e65e1a2963c72c297690053cbff78a786dd78cac2c648d217bcf2bfa1335e"
2560
+ },
2561
+ {
2562
+ "path": "experiments\\build_repro_manifest.py",
2563
+ "bytes": 8032,
2564
+ "sha256": "c5a666f1235bdca181cfbff661c031e14f9d3559b53831837e6b5b68805b67f0"
2565
+ },
2566
+ {
2567
+ "path": "experiments\\cost_guard.py",
2568
+ "bytes": 6611,
2569
+ "sha256": "35716a96fa45e7f35aa263cb3bb81b8ab4b734ae119553932e908189abce1050"
2570
+ },
2571
+ {
2572
+ "path": "experiments\\hf_budget_proxy.py",
2573
+ "bytes": 5436,
2574
+ "sha256": "ee538e89d5e732ead9aa2143840d7c3229962038a248de1fb484806a0d3b2d42"
2575
+ },
2576
+ {
2577
+ "path": "experiments\\hf_canary.py",
2578
+ "bytes": 2105,
2579
+ "sha256": "25e3b062ff937625187be731a77874551ae1780754fc1bacfe67f842788e5e35"
2580
+ },
2581
+ {
2582
+ "path": "experiments\\hf_proxy_canary.py",
2583
+ "bytes": 1082,
2584
+ "sha256": "5fab81b0153b688e6cbb8fd97b21ca4ca076bbc3f44ec7ae544ca29c0a0aba81"
2585
+ },
2586
+ {
2587
+ "path": "experiments\\pack_reproduction_bundle.py",
2588
+ "bytes": 3182,
2589
+ "sha256": "6183e401f7e22ade3c0e95c0520ed6897fdb83e2c0b850d1ce509e7b6f0693f0"
2590
+ },
2591
+ {
2592
+ "path": "experiments\\probe_dataset.py",
2593
+ "bytes": 2992,
2594
+ "sha256": "6337898ef35d1471861eb43bf1a20df936f11799246035fcb4c2d351acadd766"
2595
+ },
2596
+ {
2597
+ "path": "experiments\\probe_taxonomy_runtime.py",
2598
+ "bytes": 4460,
2599
+ "sha256": "fa901afe9154ec635cb1c2ddc9c8078c07a8f4e637f0df52b0107acc3232ac74"
2600
+ },
2601
+ {
2602
+ "path": "experiments\\run_feedback_effect.py",
2603
+ "bytes": 7923,
2604
+ "sha256": "2bd4b4e6a483d9030125f9b0c318c2aa670ae36e168113266bca3faf5380b4e8"
2605
+ },
2606
+ {
2607
+ "path": "experiments\\run_scaled_offpolicy.py",
2608
+ "bytes": 11448,
2609
+ "sha256": "59ae472e5f58d5dae2acd627d4a3d9a1dca45c1a0300ac6bdc1a27ca5f711d62"
2610
+ },
2611
+ {
2612
+ "path": "paper.pdf",
2613
+ "bytes": 1585440,
2614
+ "sha256": "00b2aee7c582f53dadf1c3b5bf6cd7fbf3da242d1a0328edb1338b9aeb318bf8"
2615
+ }
2616
+ ]
2617
+ }
2618
+ ````