Register qwen2.5-0.5b/base
Browse files- model_catalog.json +28 -13
model_catalog.json
CHANGED
|
@@ -12,11 +12,11 @@
|
|
| 12 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 13 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 14 |
"epochs": 1,
|
| 15 |
-
"lr":
|
| 16 |
"batch_size": 16,
|
| 17 |
"seed": 0,
|
| 18 |
"n_params": 494032768,
|
| 19 |
-
"timestamp": "2026-04-
|
| 20 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed0/config.json"
|
| 21 |
},
|
| 22 |
{
|
|
@@ -32,11 +32,11 @@
|
|
| 32 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 33 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 34 |
"epochs": 1,
|
| 35 |
-
"lr":
|
| 36 |
"batch_size": 16,
|
| 37 |
"seed": 1,
|
| 38 |
"n_params": 494032768,
|
| 39 |
-
"timestamp": "2026-04-
|
| 40 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed1/config.json"
|
| 41 |
},
|
| 42 |
{
|
|
@@ -52,11 +52,11 @@
|
|
| 52 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 53 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 54 |
"epochs": 1,
|
| 55 |
-
"lr":
|
| 56 |
"batch_size": 16,
|
| 57 |
"seed": 2,
|
| 58 |
"n_params": 494032768,
|
| 59 |
-
"timestamp": "2026-04-
|
| 60 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed2/config.json"
|
| 61 |
},
|
| 62 |
{
|
|
@@ -72,11 +72,11 @@
|
|
| 72 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 73 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 74 |
"epochs": 1,
|
| 75 |
-
"lr":
|
| 76 |
"batch_size": 16,
|
| 77 |
"seed": 3,
|
| 78 |
"n_params": 494032768,
|
| 79 |
-
"timestamp": "2026-04-
|
| 80 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed3/config.json"
|
| 81 |
},
|
| 82 |
{
|
|
@@ -92,11 +92,11 @@
|
|
| 92 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 93 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 94 |
"epochs": 1,
|
| 95 |
-
"lr":
|
| 96 |
"batch_size": 16,
|
| 97 |
"seed": 4,
|
| 98 |
"n_params": 494032768,
|
| 99 |
-
"timestamp": "2026-04-
|
| 100 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed4/config.json"
|
| 101 |
},
|
| 102 |
{
|
|
@@ -112,11 +112,11 @@
|
|
| 112 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 113 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 114 |
"epochs": 1,
|
| 115 |
-
"lr":
|
| 116 |
"batch_size": 16,
|
| 117 |
"seed": 5,
|
| 118 |
"n_params": 494032768,
|
| 119 |
-
"timestamp": "2026-04-
|
| 120 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed5/config.json"
|
| 121 |
},
|
| 122 |
{
|
|
@@ -126,6 +126,21 @@
|
|
| 126 |
"mode": "base",
|
| 127 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 128 |
"n_params": 494032768,
|
| 129 |
-
"timestamp": "2026-04-25T17:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
}
|
| 131 |
]
|
|
|
|
| 12 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 13 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 14 |
"epochs": 1,
|
| 15 |
+
"lr": 0.0002,
|
| 16 |
"batch_size": 16,
|
| 17 |
"seed": 0,
|
| 18 |
"n_params": 494032768,
|
| 19 |
+
"timestamp": "2026-04-25T17:40:21.084187+00:00",
|
| 20 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed0/config.json"
|
| 21 |
},
|
| 22 |
{
|
|
|
|
| 32 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 33 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 34 |
"epochs": 1,
|
| 35 |
+
"lr": 0.0002,
|
| 36 |
"batch_size": 16,
|
| 37 |
"seed": 1,
|
| 38 |
"n_params": 494032768,
|
| 39 |
+
"timestamp": "2026-04-25T17:40:21.062095+00:00",
|
| 40 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed1/config.json"
|
| 41 |
},
|
| 42 |
{
|
|
|
|
| 52 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 53 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 54 |
"epochs": 1,
|
| 55 |
+
"lr": 0.0002,
|
| 56 |
"batch_size": 16,
|
| 57 |
"seed": 2,
|
| 58 |
"n_params": 494032768,
|
| 59 |
+
"timestamp": "2026-04-25T17:40:21.054540+00:00",
|
| 60 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed2/config.json"
|
| 61 |
},
|
| 62 |
{
|
|
|
|
| 72 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 73 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 74 |
"epochs": 1,
|
| 75 |
+
"lr": 0.0002,
|
| 76 |
"batch_size": 16,
|
| 77 |
"seed": 3,
|
| 78 |
"n_params": 494032768,
|
| 79 |
+
"timestamp": "2026-04-25T17:40:21.065292+00:00",
|
| 80 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed3/config.json"
|
| 81 |
},
|
| 82 |
{
|
|
|
|
| 92 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 93 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 94 |
"epochs": 1,
|
| 95 |
+
"lr": 0.0002,
|
| 96 |
"batch_size": 16,
|
| 97 |
"seed": 4,
|
| 98 |
"n_params": 494032768,
|
| 99 |
+
"timestamp": "2026-04-25T17:40:21.046221+00:00",
|
| 100 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed4/config.json"
|
| 101 |
},
|
| 102 |
{
|
|
|
|
| 112 |
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 113 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 114 |
"epochs": 1,
|
| 115 |
+
"lr": 0.0002,
|
| 116 |
"batch_size": 16,
|
| 117 |
"seed": 5,
|
| 118 |
"n_params": 494032768,
|
| 119 |
+
"timestamp": "2026-04-25T17:40:21.071683+00:00",
|
| 120 |
"config_path": "evals/math/qwen2.5-0.5b/math_1pct_seed5/config.json"
|
| 121 |
},
|
| 122 |
{
|
|
|
|
| 126 |
"mode": "base",
|
| 127 |
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 128 |
"n_params": 494032768,
|
| 129 |
+
"timestamp": "2026-04-25T17:57:31.076340+00:00"
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"name": "qwen2.5-0.5b/owt20M",
|
| 133 |
+
"status": "VALID",
|
| 134 |
+
"status_note": "",
|
| 135 |
+
"mode": "clean",
|
| 136 |
+
"train_data_manifest": "openwebtext/subset_20M_seed0.jsonl",
|
| 137 |
+
"proxy_dataset": "openwebtext/subset_20M_seed0.jsonl",
|
| 138 |
+
"base_model": "Qwen/Qwen2.5-0.5B",
|
| 139 |
+
"epochs": 1,
|
| 140 |
+
"lr": 0.0002,
|
| 141 |
+
"batch_size": 16,
|
| 142 |
+
"n_params": 494032768,
|
| 143 |
+
"timestamp": "2026-04-25T17:40:15.553926+00:00",
|
| 144 |
+
"config_path": "evals/qwen2.5-0.5b/owt20M/config.json"
|
| 145 |
}
|
| 146 |
]
|