Hugging Face's logo

Asilarkness
/
testgeniy

Text Generation
Transformers
Safetensors
English
testgeniy
causal-lm
reasoning
mathematics
logic
long-context
4k-context
small-language-model
Model card Files
xet
 Community
testgeniy

Commit History

  1. Update handoff with on-policy error-targeted reasoning SFT v4

    Asilarkness
    Verified signature
    d714e4c

  2. V4 final rejection after protected information and reasoning gates

    Asilarkness
    Verified signature
    02335ba

  3. V4 final rejection after protected information and reasoning gates

    Asilarkness
    Verified signature
    cfe02e7

  4. V4 final rejection after protected information and reasoning gates

    Asilarkness
    Verified signature
    4086306

  5. V4 final rejection after protected information and reasoning gates

    Asilarkness
    Verified signature
    5c581bc

  6. V4 recoverable on-policy error-targeted SFT source

    Asilarkness
    Verified signature
    bce1e59

  7. V4 recoverable on-policy error-targeted SFT source

    Asilarkness
    Verified signature
    abffa02

  8. V4 error-targeted SFT dataset and leader anchoring

    Asilarkness
    Verified signature
    2fa2fce

  9. V4 error-targeted SFT dataset and leader anchoring

    Asilarkness
    Verified signature
    73c67bd

  10. V4 error-targeted SFT dataset and leader anchoring

    Asilarkness
    Verified signature
    b7bbfd9

  11. V4 scored on-policy frontier and verified minimal corrections

    Asilarkness
    Verified signature
    b69d06e

  12. V4 scored on-policy frontier and verified minimal corrections

    Asilarkness
    Verified signature
    ec5191b

  13. V4 scored on-policy frontier and verified minimal corrections

    Asilarkness
    Verified signature
    5c67cc7

  14. V4 scored on-policy frontier and verified minimal corrections

    Asilarkness
    Verified signature
    516be00

  15. Complete V4 on-policy rollout generation 300 problems / 600 trajectories

    Asilarkness
    Verified signature
    d03b5ee

  16. V4 on-policy rollout checkpoint 250/300

    Asilarkness
    Verified signature
    bb56236

  17. V4 on-policy rollout checkpoint 200/300

    Asilarkness
    Verified signature
    efc4c96

  18. V4 on-policy rollout checkpoint 150/300

    Asilarkness
    Verified signature
    9b5307c

  19. V4 on-policy rollout checkpoint 100/300

    Asilarkness
    Verified signature
    1f3ed65

  20. V4 on-policy rollout checkpoint 50/300

    Asilarkness
    Verified signature
    9c9c328

  21. Bootstrap on-policy error-targeted reasoning v4

    Asilarkness
    Verified signature
    3093e07

  22. Bootstrap on-policy error-targeted reasoning v4

    Asilarkness
    Verified signature
    2912c18

  23. Bootstrap on-policy error-targeted reasoning v4

    Asilarkness
    Verified signature
    4087ed6

  24. Confirm v3 rejection with extended fixed-MATH alpha line search

    Asilarkness
    Verified signature
    6d5909a

  25. Confirm v3 rejection with extended fixed-MATH alpha line search

    Asilarkness
    Verified signature
    d757bd2

  26. Confirm v3 rejection with extended fixed-MATH alpha line search

    Asilarkness
    Verified signature
    ef18aa4

  27. Update handoff with long reasoning SFT before RL v3

    Asilarkness
    Verified signature
    b0a1065

  28. Add v3 evaluation artifact v3_reasoning_sft_manifest.json

    Asilarkness
    Verified signature
    4ebb47e

  29. Add v3 evaluation artifact v3_fixed_scores.json

    Asilarkness
    Verified signature
    24faea4

  30. Add v3 evaluation artifact v3_math_dev_scores.json

    Asilarkness
    Verified signature
    7b717c3

  31. Add v3 evaluation artifact v3_info_dev.json

    Asilarkness
    Verified signature
    8bda0df

  32. Add v3 evaluation artifact V3_FINAL_DECISION.json

    Asilarkness
    Verified signature
    d9b05d6

  33. Reasoning SFT then RL v3: final Stage A rejection report

    Asilarkness
    Verified signature
    bc51733

  34. Add reasoning SFT v3 encoding manifest

    Asilarkness
    Verified signature
    28ff334

  35. Add reasoning SFT v3 training report

    Asilarkness
    Verified signature
    027b0e3

  36. Reasoning SFT v3: recoverable long verified SFT source

    Asilarkness
    Verified signature
    90fe6d3

  37. Add reasoning SFT then constrained RL v3 policy

    Asilarkness
    Verified signature
    9be1e6e

  38. Add reasoning SFT v3 data manifest

    Asilarkness
    Verified signature
    159a233

  39. Add 208-row held-out verified math calibration split

    Asilarkness
    Verified signature
    c51911e

  40. Bootstrap reasoning SFT v3 with 800 verified long math solutions

    Asilarkness
    Verified signature
    80f0edf

  41. Update handoff with math-augmented constrained information RL v2

    Asilarkness
    Verified signature
    a833fe9

  42. Math-augmented constrained RL v2: final rejection after fixed no-regression gate

    Asilarkness
    Verified signature
    a29f14f

  43. Math-augmented RL v2: constrained joint source with gradient projection

    Asilarkness
    Verified signature
    014c05f

  44. Math-augmented RL v2: add leader log-probabilities for joint constrained training

    Asilarkness
    Verified signature
    fe54262

  45. Math-augmented RL v2: add 96 scored math groups and symbolic rewards

    Asilarkness
    Verified signature
    9a568f8

  46. Bootstrap math-augmented information RL v2 with 96 verified decontaminated math rows

    Asilarkness
    Verified signature
    99c70bb

  47. Update Budgie handoff with information-management RL v1 result

    Asilarkness
    Verified signature
    83641cc

  48. Information RL v1 hygiene: remove superseded and rejected full checkpoints

    Asilarkness
    Verified signature
    980a3df

  49. Information-management RL v1: final rejection after protected reasoning gate

    Asilarkness
    Verified signature
    efa1162

  50. Information RL v1: preliminary alpha-0.025 candidate and held-out decision

    Asilarkness
    Verified signature
    d53a83f