Skip to content

Commit 76b5d0b

Browse files
committed
Fix outputs documentation with correct parameter syntax
Correct the --outputs parameter examples to show comma-separated values instead of space-separated, which doesn't work in all environments. The parameter supports comma-separated format aliases or multiple flags. Signed-off-by: Maryam Tahhan <[email protected]>
1 parent adfa108 commit 76b5d0b

File tree

1 file changed

+62
-2
lines changed

1 file changed

+62
-2
lines changed

docs/guides/outputs.md

Lines changed: 62 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -66,16 +66,76 @@ GuideLLM supports saving benchmark results to files in various formats, includin
6666
- **Output Formats**: Use the `--outputs` argument to specify which formats or exact file names (with supported file extensions, e.g. `benchmarks.json`) to generate. By default, JSON, CSV, and HTML are generated.
6767
- **Sampling**: To limit the size of the output files and number of detailed request samples included, you can configure sampling options using the `--sample-requests` argument.
6868

69-
Example command to save results in specific formats:
69+
#### Example commands to save results in specific formats:
7070

71+
The `--outputs` parameter accepts output formats in the following ways:
72+
73+
**Command Line:**
74+
```bash
75+
# Comma-separated format aliases (recommended)
76+
--outputs json,csv,html
77+
78+
# Or using multiple flags
79+
--outputs json --outputs csv --outputs html
80+
```
81+
82+
**Environment Variables:**
83+
```bash
84+
# Comma-separated values for environment variables
85+
-e GUIDELLM_OUTPUTS=json,csv,html
86+
```
87+
88+
#### Examples
89+
90+
**Example 1: Comma-separated format aliases (recommended)**
91+
```bash
92+
guidellm benchmark \
93+
--target "http://localhost:8000" \
94+
--profile sweep \
95+
--max-seconds 30 \
96+
--data "prompt_tokens=256,output_tokens=128" \
97+
--output-dir "results/" \
98+
--outputs json,csv,html \
99+
--sample-requests 20
100+
```
101+
102+
**Example 2: Multiple flags for each format**
103+
```bash
104+
guidellm benchmark \
105+
--target "http://localhost:8000" \
106+
--profile sweep \
107+
--max-seconds 30 \
108+
--data "prompt_tokens=256,output_tokens=128" \
109+
--output-dir "results/" \
110+
--outputs json \
111+
--outputs csv \
112+
--outputs html \
113+
--sample-requests 20
114+
```
115+
116+
**Example 3: Using environment variables (Docker/Podman)**
117+
```bash
118+
podman run --rm -it --network=host \
119+
-v "/tmp/results:/results:z" \
120+
-e GUIDELLM_TARGET=http://localhost:8000 \
121+
-e GUIDELLM_PROFILE=sweep \
122+
-e GUIDELLM_MAX_SECONDS=30 \
123+
-e GUIDELLM_DATA="prompt_tokens=256,output_tokens=128" \
124+
-e GUIDELLM_OUTPUT_DIR=/results \
125+
-e GUIDELLM_OUTPUTS=json,csv,html \
126+
-e GUIDELLM_SAMPLE_REQUESTS=20 \
127+
ghcr.io/vllm-project/guidellm:latest
128+
```
129+
130+
**Example 4: Single output format**
71131
```bash
72132
guidellm benchmark \
73133
--target "http://localhost:8000" \
74134
--profile sweep \
75135
--max-seconds 30 \
76136
--data "prompt_tokens=256,output_tokens=128" \
77137
--output-dir "results/" \
78-
--outputs json csv \
138+
--outputs json \
79139
--sample-requests 20
80140
```
81141

0 commit comments

Comments
 (0)