Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add run entries for Lite with instructions #3206

Merged
merged 2 commits into from
Dec 10, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 11 additions & 1 deletion docs/reproducing_leaderboards.md
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ helm-server --suite $SUITE_NAME

The following specifies the appropriate parameters and configuration files for a leaderboard, given its project and version number.

### Lite
### Lite for non-instruction-following models

```bash
export RUN_ENTRIES_CONF_PATH=run_entries_lite_20240424.conf
Expand All @@ -44,6 +44,16 @@ export MAX_EVAL_INSTANCES=1000
export PRIORITY=2
```

### Lite for instruction-following models

```bash
export RUN_ENTRIES_CONF_PATH=run_entries_lite_20240424_output_format_instructions.conf
export SCHEMA_PATH=schema_lite.yaml
export NUM_TRAIN_TRIALS=1
export MAX_EVAL_INSTANCES=1000
export PRIORITY=2
```

### Classic before v0.2.4

```bash
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,50 @@
# HELM scenarios.

entries: [
# NarrativeQA
{description: "narrative_qa:model=text,output_format_instructions=narrative_qa", priority: 1}

# NaturalQuestions
{description: "natural_qa:model=text,mode=openbook_longans,output_format_instructions=natural_qa", priority: 1}
{description: "natural_qa:model=text,mode=closedbook,output_format_instructions=natural_qa", priority: 1}

# OpenbookQA
{description: "commonsense:model=text_code,dataset=openbookqa,method=multiple_choice_joint,output_format_instructions=openbookqa", priority: 1}

# MMLU
{description: "mmlu:model=text,subject=abstract_algebra,output_format_instructions=mmlu", priority: 2}
{description: "mmlu:model=text,subject=college_chemistry,output_format_instructions=mmlu", priority: 2}
{description: "mmlu:model=text,subject=computer_security,output_format_instructions=mmlu", priority: 2}
{description: "mmlu:model=text,subject=econometrics,output_format_instructions=mmlu", priority: 2}
{description: "mmlu:model=text,subject=us_foreign_policy,output_format_instructions=mmlu", priority: 2}

# MATH
{description: "math:model=text_code,subject=number_theory,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=intermediate_algebra,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=algebra,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=prealgebra,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=geometry,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=counting_and_probability,level=1,use_chain_of_thought=True", priority: 2}
{description: "math:model=text_code,subject=precalculus,level=1,use_chain_of_thought=True", priority: 2}

# GSM
{description: "gsm:model=text_code,stop=none", priority: 2}

# LegalBench
{description: "legalbench:model=text_code,subset=abercrombie,output_format_instructions=legalbench", priority: 2}
{description: "legalbench:model=text_code,subset=corporate_lobbying,output_format_instructions=legalbench", priority: 2}
{description: "legalbench:model=text_code,subset=international_citizenship_questions,output_format_instructions=legalbench", priority: 2}
{description: "legalbench:model=text_code,subset=function_of_decision_section,output_format_instructions=legalbench", priority: 2}
{description: "legalbench:model=text_code,subset=proa,output_format_instructions=legalbench", priority: 2}

# MedQA
{description: "med_qa:model=text_code,output_format_instructions=med_qa", priority: 2}

# WMT14
{description: "wmt_14:language_pair=cs-en,model=text,output_format_instructions=wmt_14", priority: 2}
{description: "wmt_14:language_pair=de-en,model=text,output_format_instructions=wmt_14", priority: 2}
{description: "wmt_14:language_pair=fr-en,model=text,output_format_instructions=wmt_14", priority: 2}
{description: "wmt_14:language_pair=hi-en,model=text,output_format_instructions=wmt_14", priority: 2}
{description: "wmt_14:language_pair=ru-en,model=text,output_format_instructions=wmt_14", priority: 2}
]

Loading