docs: center fev in model comparison
Browse files- MODEL_CARD.md +3 -3
- README.md +3 -3
- manifest.json +4 -4
MODEL_CARD.md
CHANGED
|
@@ -114,10 +114,10 @@ Checkpoint and adapter files retain their original precision and hashes. Exact a
|
|
| 114 |
|
| 115 |
The new common native typed-decision comparison against pinned Kev-4B and Jev is documented in [BENCHMARKS.md](BENCHMARKS.md), with separate public-development and authored-Russian denominators, probability metrics and reproducibility commands. It is distinct from the routing-composition evidence below.
|
| 116 |
|
| 117 |
-
| Native comparison |
|
| 118 |
|---|---:|---:|---:|
|
| 119 |
-
| Public development |
|
| 120 |
-
| Russian probes |
|
| 121 |
|
| 122 |
Public development: 764 questions. Authored Russian probes: 144 questions.
|
| 123 |
|
|
|
|
| 114 |
|
| 115 |
The new common native typed-decision comparison against pinned Kev-4B and Jev is documented in [BENCHMARKS.md](BENCHMARKS.md), with separate public-development and authored-Russian denominators, probability metrics and reproducibility commands. It is distinct from the routing-composition evidence below.
|
| 116 |
|
| 117 |
+
| Native comparison | Kev‑4B | **Fev‑4B** | Jev‑1.14.0 |
|
| 118 |
|---|---:|---:|---:|
|
| 119 |
+
| Public development | 80.2% | **83.6%** | 84.6% |
|
| 120 |
+
| Russian probes | 99.3% | **100.0%** | 100.0% |
|
| 121 |
|
| 122 |
Public development: 764 questions. Authored Russian probes: 144 questions.
|
| 123 |
|
README.md
CHANGED
|
@@ -114,10 +114,10 @@ Checkpoint and adapter files retain their original precision and hashes. Exact a
|
|
| 114 |
|
| 115 |
The new common native typed-decision comparison against pinned Kev-4B and Jev is documented in [BENCHMARKS.md](BENCHMARKS.md), with separate public-development and authored-Russian denominators, probability metrics and reproducibility commands. It is distinct from the routing-composition evidence below.
|
| 116 |
|
| 117 |
-
| Native comparison |
|
| 118 |
|---|---:|---:|---:|
|
| 119 |
-
| Public development |
|
| 120 |
-
| Russian probes |
|
| 121 |
|
| 122 |
Public development: 764 questions. Authored Russian probes: 144 questions.
|
| 123 |
|
|
|
|
| 114 |
|
| 115 |
The new common native typed-decision comparison against pinned Kev-4B and Jev is documented in [BENCHMARKS.md](BENCHMARKS.md), with separate public-development and authored-Russian denominators, probability metrics and reproducibility commands. It is distinct from the routing-composition evidence below.
|
| 116 |
|
| 117 |
+
| Native comparison | Kev‑4B | **Fev‑4B** | Jev‑1.14.0 |
|
| 118 |
|---|---:|---:|---:|
|
| 119 |
+
| Public development | 80.2% | **83.6%** | 84.6% |
|
| 120 |
+
| Russian probes | 99.3% | **100.0%** | 100.0% |
|
| 121 |
|
| 122 |
Public development: 764 questions. Authored Russian probes: 144 questions.
|
| 123 |
|
manifest.json
CHANGED
|
@@ -27,16 +27,16 @@
|
|
| 27 |
"sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a"
|
| 28 |
},
|
| 29 |
"MODEL_CARD.md": {
|
| 30 |
-
"bytes":
|
| 31 |
-
"sha256": "
|
| 32 |
},
|
| 33 |
"NOTICE": {
|
| 34 |
"bytes": 132,
|
| 35 |
"sha256": "381fb81e814716a5ef8e5eae26a08607d052cf12f019dba9da9e81298ef5cb75"
|
| 36 |
},
|
| 37 |
"README.md": {
|
| 38 |
-
"bytes":
|
| 39 |
-
"sha256": "
|
| 40 |
},
|
| 41 |
"RUNNING.md": {
|
| 42 |
"bytes": 7334,
|
|
|
|
| 27 |
"sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a"
|
| 28 |
},
|
| 29 |
"MODEL_CARD.md": {
|
| 30 |
+
"bytes": 11706,
|
| 31 |
+
"sha256": "48e68df982a747d2ffde857f03ec16bcfb942eb80b8033e8233c050238b13961"
|
| 32 |
},
|
| 33 |
"NOTICE": {
|
| 34 |
"bytes": 132,
|
| 35 |
"sha256": "381fb81e814716a5ef8e5eae26a08607d052cf12f019dba9da9e81298ef5cb75"
|
| 36 |
},
|
| 37 |
"README.md": {
|
| 38 |
+
"bytes": 11706,
|
| 39 |
+
"sha256": "48e68df982a747d2ffde857f03ec16bcfb942eb80b8033e8233c050238b13961"
|
| 40 |
},
|
| 41 |
"RUNNING.md": {
|
| 42 |
"bytes": 7334,
|