Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
gaoqiong
lm-evaluation-harness
Commits
292fdae5
Commit
292fdae5
authored
May 07, 2024
by
Yen-Ting Lin
Browse files
Add trust_remote_code and wandb_args to harness_eval.slurm, and add run_all.sh script
parent
f4f44b96
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
46 additions
and
3 deletions
+46
-3
harness_eval.slurm
harness_eval.slurm
+3
-3
run_all.sh
run_all.sh
+43
-0
No files found.
harness_eval.slurm
View file @
292fdae5
...
@@ -53,7 +53,7 @@ echo "TASKS: $tasks"
...
@@ -53,7 +53,7 @@ echo "TASKS: $tasks"
export
PROGRAM
=
"
\
export
PROGRAM
=
"
\
lm_eval
\
lm_eval
\
--model hf
\
--model hf
\
--model_args pretrained=
$model
,parallelize=True
\
--model_args pretrained=
$model
,parallelize=True
,trust_remote_code=True
\
--tasks
$tasks
\
--tasks
$tasks
\
--num_fewshot 0
\
--num_fewshot 0
\
--batch_size 8
\
--batch_size 8
\
...
@@ -61,8 +61,8 @@ lm_eval \
...
@@ -61,8 +61,8 @@ lm_eval \
--write_out
\
--write_out
\
--log_samples
\
--log_samples
\
--verbosity DEBUG
\
--verbosity DEBUG
\
--wandb_args project=lm-eval-harness-integration,job_type=eval
\
--wandb_args project=lm-eval-harness-integration,job_type=eval
,name=
$model
\
--hf_hub_log_args
=
hub_results_org=yentinglin,hub_repo_name=
twl
lm-eval-results
\
--hf_hub_log_args
hub_results_org=yentinglin,hub_repo_name=lm-eval-results
,push_results_to_hub=True,push_samples_to_hub=True,public_repo=False
\
--seed 42
\
--seed 42
\
--trust_remote_code
\
--trust_remote_code
\
"
"
...
...
run_all.sh
0 → 100644
View file @
292fdae5
#!/bin/bash
# Define the models to run
declare
-a
models
=(
"yentinglin/Llama-3-Taiwan-70B-Instruct"
"yentinglin/Taiwan-Llama-3-70B-Cooldown"
"yentinglin/Taiwan-Llama-3-70B"
"yentinglin/Taiwan-Llama-3-8B-Instruct"
"yentinglin/Taiwan-Llama-3-8B-Cooldown"
"yentinglin/Taiwan-Llama-3-8B"
"meta-llama/Meta-Llama-3-70B-Instruct"
"meta-llama/Meta-Llama-3-70B"
"meta-llama/Meta-Llama-3-8B-Instruct"
"meta-llama/Meta-Llama-3-8B"
"Qwen/Qwen1.5-110B-Chat"
"Qwen/Qwen1.5-72B-Chat"
"deepseek-ai/DeepSeek-V2-Chat"
"01-ai/Yi-34B-Chat"
"CohereForAI/c4ai-command-r-plus"
"mistralai/Mixtral-8x22B-Instruct-v0.1"
"MediaTek-Research/Breeze-7B-Instruct-v1_0"
"MediaTek-Research/Breeze-7B-Base-v1_0"
"taide/Llama3-TAIDE-LX-8B-Chat-Alpha1"
"taide/TAIDE-LX-7B-Chat"
"taide/TAIDE-LX-7B"
"microsoft/Phi-3-mini-4k-instruct"
"apple/OpenELM-3B-Instruct"
)
# SLURM script to be used
SLURM_SCRIPT
=
"harness_eval.slurm"
# Parameters for the script
PARAMS
=
"tmlu,twllm_eval,tw_legal,ccp,pega,tmmluplus"
# Loop through each model and submit a job
for
model
in
"
${
models
[@]
}
"
do
echo
"Submitting job for
$model
"
sbatch
$SLURM_SCRIPT
$model
$PARAMS
done
echo
"All jobs submitted"
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment