wadhma commited on
Commit
4affe1f
·
verified ·
1 Parent(s): 49f9b28

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +59 -0
README.md CHANGED
@@ -4,6 +4,10 @@ configs:
4
  data_files:
5
  - split: train
6
  path: data/rat.jsonl
 
 
 
 
7
  ---
8
  <!-- config:rat -->
9
  ## rat
@@ -69,3 +73,58 @@ Full run manifest: manifests/rat.json.
69
  | ts | unix time the row was written |
70
 
71
  <!-- /config:rat -->
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  data_files:
5
  - split: train
6
  path: data/rat.jsonl
7
+ - config_name: simpleqa_verified
8
+ data_files:
9
+ - split: train
10
+ path: data/simpleqa_verified.jsonl
11
  ---
12
  <!-- config:rat -->
13
  ## rat
 
73
  | ts | unix time the row was written |
74
 
75
  <!-- /config:rat -->
76
+
77
+ <!-- config:simpleqa_verified -->
78
+ ## simpleqa_verified
79
+
80
+ 20000 rows (20000 successful, 0 failed). finish_reason counts: {'stop': 19990, 'length': 10}.
81
+ Model: Qwen/Qwen3-8B (revision b968826d9c46dd6066d109eabc6255188de91218), thinking on.
82
+
83
+ ### Sampling (as sent to the server)
84
+ ```json
85
+ {
86
+ "temperature": 0.9,
87
+ "top_p": 0.95,
88
+ "top_k": 20,
89
+ "min_p": 0.0,
90
+ "max_tokens": 32768,
91
+ "seed": "not sent"
92
+ }
93
+ ```
94
+
95
+ ### Runs
96
+ git sha(s): ae98fd5044474596e571b2f0f8bd8723e3f57335;
97
+ slurm job(s): 19358602.
98
+ Full run manifest: manifests/simpleqa_verified.json.
99
+
100
+ ### Dataset provenance
101
+ ```json
102
+ {
103
+ "dataset": "simpleqa_verified",
104
+ "repo": "google/simpleqa-verified",
105
+ "file": "simpleqa_verified.csv",
106
+ "revision": "0dc97e0d28d8233463e005cdc4475cc2a13ba2dc",
107
+ "sha256": "b5db21155444763543fe31b67e7cf28ce2bb225742a5b889421c5f182e2f92f5",
108
+ "rows_total": 1000,
109
+ "instruction": null
110
+ }
111
+ ```
112
+
113
+ ### Columns
114
+ | Column | Description |
115
+ |---|---|
116
+ | dataset | dataset name (also the config name of this subset) |
117
+ | instance_id | instance id within the dataset |
118
+ | sample_idx | index of this sample for the instance |
119
+ | prompt_sha1 | sha1 of the exact prompt string sent to the server |
120
+ | prompt_tokens | prompt length in tokens |
121
+ | content | raw model output, untouched (thinking and </think> inline) |
122
+ | finish_reason | vLLM finish_reason: stop or length |
123
+ | usage | vLLM usage block: prompt_tokens, completion_tokens, total_tokens |
124
+ | success | true if the request returned a generation |
125
+ | error | error string for a failed request, else null |
126
+ | attempts | requests made for this row (1 + retries) |
127
+ | latency_s | wall seconds for this request, from client send to response |
128
+ | ts | unix time the row was written |
129
+
130
+ <!-- /config:simpleqa_verified -->