EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Accuracy report formatting (#2279)

Browse Source 
### What this PR does / why we need it?
Accuracy report formatting

### Does this PR introduce _any_ user-facing change?
no

### How was this patch tested?
CI passed with existing test.


- vLLM version: v0.10.0
- vLLM main:
53415653ff

---------

Signed-off-by: Icey <1790571317@qq.com>
This commit is contained in:
Icey
2025-08-25 09:39:30 +08:00
committed by GitHub
parent f796e6280b
commit 891b2bfe71
2 changed files with 15 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
tests/e2e/models/report_template.md
View File

@@ -1,24 +1,21 @@
# {{ model_name }} # {{ model_name }}
**vLLM Version**: vLLM: {{ vllm_version }} ([{{ vllm_commit[:7] }}](https://github.com/vllm-project/vllm/commit/{{ vllm_commit }})), - **vLLM Version**: vLLM: {{ vllm_version }} ([{{ vllm_commit[:7] }}](https://github.com/vllm-project/vllm/commit/{{ vllm_commit }})), **vLLM Ascend Version**: {{ vllm_ascend_version }} ([{{ vllm_ascend_commit[:7] }}](https://github.com/vllm-project/vllm-ascend/commit/{{ vllm_ascend_commit }}))
**vLLM Ascend Version**: {{ vllm_ascend_version }} ([{{ vllm_ascend_commit[:7] }}](https://github.com/vllm-project/vllm-ascend/commit/{{ vllm_ascend_commit }})) - **Software Environment**: **CANN**: {{ cann_version }}, **PyTorch**: {{ torch_version }}, **torch-npu**: {{ torch_npu_version }}
**Software Environment**: CANN: {{ cann_version }}, PyTorch: {{ torch_version }}, torch-npu: {{ torch_npu_version }} - **Hardware Environment**: Atlas A2 Series
**Hardware Environment**: Atlas A2 Series - **Parallel mode**: {{ parallel_mode }}
**Datasets**: {{ datasets }} - **Execution mode**: ACLGraph
**Parallel Mode**: TP
**Execution Mode**: ACLGraph
**Command**: **Command**:
```bash ```bash
export MODEL_ARGS={{ model_args }} export MODEL_ARGS={{ model_args }}
lm_eval --model {{ model_type }} --model_args $MODEL_ARGS --tasks {{ datasets }} \ lm_eval --model {{ model_type }} --model_args $MODEL_ARGS --tasks {{ datasets }} \
--apply_chat_template {{ apply_chat_template }} --fewshot_as_multiturn {{ fewshot_as_multiturn }} {% if num_fewshot is defined and num_fewshot != "N/A" %} --num_fewshot {{ num_fewshot }} {% endif %} \ {% if apply_chat_template %} --apply_chat_template {{ apply_chat_template }} {% endif %} {% if fewshot_as_multiturn %} --fewshot_as_multiturn {{ fewshot_as_multiturn }} {% endif %} {% if num_fewshot is defined and num_fewshot != "N/A" %} --num_fewshot {{ num_fewshot }} {% endif %} {% if limit is defined and limit != "N/A" %} --limit {{ limit }} {% endif %} --batch_size {{ batch_size}}
--limit {{ limit }} --batch_size {{ batch_size}}
``` ```
| Task | Metric | Value | Stderr | | Task | Metric | Value | Stderr |
|-----------------------|-------------|----------:|-------:| |-----------------------|-------------|----------:|-------:|
{% for row in rows -%} {% for row in rows -%}
| {{ row.task.rjust(23) }} | {{ row.metric.rjust(15) }} |{{ row.value }} | ± {{ "%.4f" | format(row.stderr | float) }} | | {{ row.task }} | {{ row.metric }} | {{ row.value }} | ± {{ "%.4f" | format(row.stderr | float) }} |
{% endfor %} {% endfor %}

11
tests/e2e/models/test_lm_eval_correctness.py
View File

@@ -65,6 +65,10 @@ def generate_report(tp_size, eval_config, report_data, report_dir, env_config):
template = env.get_template("report_template.md") template = env.get_template("report_template.md")
model_args = build_model_args(eval_config, tp_size) model_args = build_model_args(eval_config, tp_size)
parallel_mode = f"TP{model_args.get('tensor_parallel_size', 1)}"
if model_args.get('enable_expert_parallel', False):
parallel_mode += " + EP"
report_content = template.render( report_content = template.render(
vllm_version=env_config.vllm_version, vllm_version=env_config.vllm_version,
vllm_commit=env_config.vllm_commit, vllm_commit=env_config.vllm_commit,
@@ -79,10 +83,11 @@ def generate_report(tp_size, eval_config, report_data, report_dir, env_config):
datasets=",".join([task["name"] for task in eval_config["tasks"]]), datasets=",".join([task["name"] for task in eval_config["tasks"]]),
apply_chat_template=eval_config.get("apply_chat_template", True), apply_chat_template=eval_config.get("apply_chat_template", True),
fewshot_as_multiturn=eval_config.get("fewshot_as_multiturn", True), fewshot_as_multiturn=eval_config.get("fewshot_as_multiturn", True),
limit=eval_config.get("limit", None), limit=eval_config.get("limit", "N/A"),
batch_size="auto", batch_size="auto",
num_fewshot=eval_config.get("num_fewshot", "N/A"), num_fewshot=eval_config.get("num_fewshot", "N/A"),
rows=report_data["rows"]) rows=report_data["rows"],
parallel_mode=parallel_mode)
report_output = os.path.join( report_output = os.path.join(
report_dir, f"{os.path.basename(eval_config['model_name'])}.md") report_dir, f"{os.path.basename(eval_config['model_name'])}.md")
@@ -123,7 +128,7 @@ def test_lm_eval_correctness_param(config_filename, tp_size, report_dir,
for metric in task["metrics"]: for metric in task["metrics"]:
metric_name = metric["name"] metric_name = metric["name"]
ground_truth = metric["value"] ground_truth = metric["value"]
measured_value = task_result[metric_name] measured_value = round(task_result[metric_name], 4)
task_success = bool( task_success = bool(
np.isclose(ground_truth, measured_value, rtol=RTOL)) np.isclose(ground_truth, measured_value, rtol=RTOL))
success = success and task_success success = success and task_success