itzsuni chetangadgil commited on
Commit
14b48c9
·
0 Parent(s):

Duplicate from metrum-ai/llm-perfdata

Browse files

Co-authored-by: Chetan Gadgil <chetangadgil@users.noreply.huggingface.co>

Files changed (3) hide show
  1. .gitattributes +59 -0
  2. README.md +142 -0
  3. data/train-00000-of-00001.parquet +3 -0
.gitattributes ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.lz4 filter=lfs diff=lfs merge=lfs -text
12
+ *.mds filter=lfs diff=lfs merge=lfs -text
13
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
14
+ *.model filter=lfs diff=lfs merge=lfs -text
15
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
16
+ *.npy filter=lfs diff=lfs merge=lfs -text
17
+ *.npz filter=lfs diff=lfs merge=lfs -text
18
+ *.onnx filter=lfs diff=lfs merge=lfs -text
19
+ *.ot filter=lfs diff=lfs merge=lfs -text
20
+ *.parquet filter=lfs diff=lfs merge=lfs -text
21
+ *.pb filter=lfs diff=lfs merge=lfs -text
22
+ *.pickle filter=lfs diff=lfs merge=lfs -text
23
+ *.pkl filter=lfs diff=lfs merge=lfs -text
24
+ *.pt filter=lfs diff=lfs merge=lfs -text
25
+ *.pth filter=lfs diff=lfs merge=lfs -text
26
+ *.rar filter=lfs diff=lfs merge=lfs -text
27
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
28
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
29
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
30
+ *.tar filter=lfs diff=lfs merge=lfs -text
31
+ *.tflite filter=lfs diff=lfs merge=lfs -text
32
+ *.tgz filter=lfs diff=lfs merge=lfs -text
33
+ *.wasm filter=lfs diff=lfs merge=lfs -text
34
+ *.xz filter=lfs diff=lfs merge=lfs -text
35
+ *.zip filter=lfs diff=lfs merge=lfs -text
36
+ *.zst filter=lfs diff=lfs merge=lfs -text
37
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
38
+ # Audio files - uncompressed
39
+ *.pcm filter=lfs diff=lfs merge=lfs -text
40
+ *.sam filter=lfs diff=lfs merge=lfs -text
41
+ *.raw filter=lfs diff=lfs merge=lfs -text
42
+ # Audio files - compressed
43
+ *.aac filter=lfs diff=lfs merge=lfs -text
44
+ *.flac filter=lfs diff=lfs merge=lfs -text
45
+ *.mp3 filter=lfs diff=lfs merge=lfs -text
46
+ *.ogg filter=lfs diff=lfs merge=lfs -text
47
+ *.wav filter=lfs diff=lfs merge=lfs -text
48
+ # Image files - uncompressed
49
+ *.bmp filter=lfs diff=lfs merge=lfs -text
50
+ *.gif filter=lfs diff=lfs merge=lfs -text
51
+ *.png filter=lfs diff=lfs merge=lfs -text
52
+ *.tiff filter=lfs diff=lfs merge=lfs -text
53
+ # Image files - compressed
54
+ *.jpg filter=lfs diff=lfs merge=lfs -text
55
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
56
+ *.webp filter=lfs diff=lfs merge=lfs -text
57
+ # Video files - compressed
58
+ *.mp4 filter=lfs diff=lfs merge=lfs -text
59
+ *.webm filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,142 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ task_categories:
5
+ - other
6
+ tags:
7
+ - llm
8
+ - performance
9
+ - benchmarking
10
+ - throughput
11
+ - latency
12
+ - hardware
13
+ license: mit
14
+ ---
15
+
16
+ # Dataset Card: llm-perfdata
17
+
18
+ ## Dataset Description
19
+
20
+ This dataset curates throughput and latency benchmarks for popular large language models across hardware targets. Each row represents an observed configuration—model, precision, serving engine, and load profile—paired with sources that document how the measurement was collected. The goal is to keep a transparent, reproducible ledger that helps compare serving trade-offs without digging through scattered notebooks.
21
+
22
+ ### Provenance & Caveats
23
+
24
+ All entries are derived solely from online, publicly available sources. Because performance numbers depend on external documentation, there may be gaps, inconsistencies, or occasional inaccuracies. Expect the dataset to drift out of date as serving stacks and software releases evolve; refresh measurements regularly when citing results.
25
+
26
+ ## Data Schema
27
+
28
+ The dataset contains the following columns:
29
+
30
+ - **Model** — published model identifier (e.g., `DeepSeek-R1-Distill-Qwen`).
31
+ - **Size** — parameter scale shorthand such as `7B` or `32B`.
32
+ - **Precision** — numeric precision used during serving (`FP16`, `BF16`, `INT4`, etc.).
33
+ - **GPU_Type** — accelerator family (for example `NVIDIA A100`).
34
+ - **Num_GPUs** — integer count of GPUs participating in the run.
35
+ - **Serving_Engine** — runtime layer (`vLLM`, `TensorRT-LLM`, custom stacks).
36
+ - **Concurrency** — concurrent request count exercised in the benchmark.
37
+ - **Tokens_per_sec** — aggregate output throughput.
38
+ - **TTFT_ms** — time-to-first-token in milliseconds.
39
+ - **TPOT_ms** — tail period of token generation in milliseconds.
40
+ - **Prompt_Tokens** / **Output_Tokens** — tokens in the input and generated output.
41
+ - **Context_Window** — maximum supported tokens for the configuration.
42
+ - **Quantization** — applied quantization strategy, if any.
43
+ - **Source_URL** — public link to the benchmark report or raw logs.
44
+ - **Source_Notes** — short free-text context, hardware topology, or caveats.
45
+
46
+ Leave optional numeric metrics blank when a source does not provide them and describe missing context in `Source_Notes`.
47
+
48
+ ## Usage
49
+
50
+ ```python
51
+ from datasets import load_dataset
52
+
53
+ dataset = load_dataset("metrum-ai/llm-perfdata")
54
+ print(dataset)
55
+
56
+ # Access the data
57
+ for example in dataset['train']:
58
+ print(f"Model: {example['Model']}")
59
+ print(f"Throughput: {example['Tokens_per_sec']} tokens/sec")
60
+ print(f"Source: {example['Source_URL']}")
61
+ ```
62
+
63
+ Or load directly as a pandas DataFrame:
64
+
65
+ ```python
66
+ import pandas as pd
67
+ from datasets import load_dataset
68
+
69
+ dataset = load_dataset("metrum-ai/llm-perfdata")
70
+ df = dataset['train'].to_pandas()
71
+
72
+ # Filter by model and precision
73
+ filtered = df[(df["Model"] == "DeepSeek-R1-Distill-Qwen") & (df["Precision"] == "FP16")]
74
+ ```
75
+
76
+ Analysts typically pivot on `Serving_Engine` and `Concurrency` to compare throughput scaling. Cite the `Source_URL` when referencing numbers externally.
77
+
78
+ ## Attribution
79
+
80
+ **If you use this dataset, you must provide attribution to Metrum AI.** Please cite this dataset using the citation format provided below.
81
+
82
+ ## License
83
+
84
+ This dataset is released under the MIT License.
85
+
86
+ **MIT License**
87
+
88
+ Copyright (c) 2025 Metrum AI
89
+
90
+ Permission is hereby granted, free of charge, to any person obtaining a copy
91
+ of this software and/or data and associated documentation files (the "Software and/or Data"), to deal
92
+ in the Software and/or Data without restriction, including without limitation the rights
93
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
94
+ copies of the Software and/or Data, and to permit persons to whom the Software and/or Data is
95
+ furnished to do so, subject to the following conditions:
96
+
97
+ The above copyright notice and this permission notice shall be included in all
98
+ copies or substantial portions of the Software and/or Data.
99
+
100
+ **THE SOFTWARE AND/OR DATA IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
101
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
102
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
103
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
104
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
105
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE AND/OR DATA OR THE USE OR OTHER DEALINGS IN THE
106
+ SOFTWARE AND/OR DATA.**
107
+
108
+ ### No Warranty and Limitation of Liability
109
+
110
+ **THIS DATASET IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED.** The data is compiled from publicly available sources and may contain errors, inaccuracies, or become outdated. Metrum AI makes no representations or warranties regarding the accuracy, completeness, reliability, or suitability of this dataset for any purpose.
111
+
112
+ **IN NO EVENT SHALL METRUM AI BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY**, whether in an action of contract, tort, or otherwise, arising from, out of, or in connection with this dataset or the use or other dealings in this dataset. This includes, without limitation, direct, indirect, incidental, special, consequential, or punitive damages, or any loss of profits, revenues, data, use, goodwill, or other intangible losses.
113
+
114
+ Use of this dataset is at your own risk.
115
+
116
+ ### Additional Disclaimers
117
+
118
+ **No Endorsement**: The inclusion of any model, hardware, software, or service in this dataset does not constitute an endorsement, recommendation, or approval by Metrum AI. All trademarks, product names, and company names are the property of their respective owners.
119
+
120
+ **Third-Party Sources**: This dataset aggregates data from publicly available third-party sources. Metrum AI does not control, verify, or guarantee the accuracy of information from these sources. Users should independently verify any information before relying on it.
121
+
122
+ **No Professional Advice**: This dataset is provided for informational and research purposes only. It does not constitute professional, technical, or business advice. Users should consult with qualified professionals for decisions based on this data.
123
+
124
+ **Data Completeness**: This dataset may not include all available performance benchmarks. The absence of data for a particular model, hardware configuration, or metric does not imply that such data does not exist or is not relevant.
125
+
126
+ **No Guarantee of Availability**: Metrum AI does not guarantee that this dataset will be available at all times or that it will be updated regularly. The dataset may be modified, discontinued, or removed without notice.
127
+
128
+ **Forward-Looking Statements**: Any performance metrics or benchmarks in this dataset reflect historical or current conditions and may not be indicative of future performance.
129
+
130
+ **User Responsibility**: Users are solely responsible for their use of this dataset, including compliance with applicable laws, regulations, and third-party rights. Users should conduct their own due diligence before making any decisions based on this data.
131
+
132
+ ## Citation
133
+
134
+ ```bibtex
135
+ @dataset{llm_perfdata,
136
+ title = {LLM Perfdata},
137
+ author = {Metrum AI},
138
+ year = {2025},
139
+ publisher = {Hugging Face},
140
+ url = {https://huggingface.co/datasets/metrum-ai/llm-perfdata}
141
+ }
142
+ ```
data/train-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e42a8a8c7711db26e75a0383cf2a0edd84e39d566b110859e4cf3fa6ebfb834
3
+ size 11252