diff --git a/.github/instructions/.instructions.md b/.github/instructions/.instructions.md deleted file mode 100644 index 3c7c3c2..0000000 --- a/.github/instructions/.instructions.md +++ /dev/null @@ -1,18 +0,0 @@ ---- -applyTo: '**' ---- -# Implementation -- When asked for a change or implementation, only work on the specific task I have given you with the most concise and elegant solution that changes as little code as possible. -- If the requested feature has unintended side effect, let the user know before implementing. - -# Review -- When reviewing, first check the code diff for potential bugs and inconsistencies. After that, check if the change would cause any bugs somewhere else in the codebase. - -# Code Style -- Follow best practices to keep code clean and succinct. - -# General -- You should give detailed, professional, and nicely formatted explanation. -- Always think slowly and carefully before doing anything. Be objective. -- If instruction is unclear or ambiguous, ask the user for clarification. - diff --git a/scripts/main_exp/eval/base_model.sh b/scripts/main_exp/eval/base_model.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/eval/cd.sh b/scripts/main_exp/eval/cd.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/eval/cd_oracle.sh b/scripts/main_exp/eval/cd_oracle.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/eval/d2l.sh b/scripts/main_exp/eval/d2l.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/eval/llmlingua.sh b/scripts/main_exp/eval/llmlingua.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/eval/t2l.sh b/scripts/main_exp/eval/t2l.sh old mode 100644 new mode 100755 diff --git a/scripts/main_exp/gen_data.sh b/scripts/main_exp/gen_data.sh old mode 100644 new mode 100755 index f2adb4f..8eced94 --- a/scripts/main_exp/gen_data.sh +++ b/scripts/main_exp/gen_data.sh @@ -3,7 +3,7 @@ uv run data/download_fineweb_edu.py # generate qa data # run from 000 to 013 -for shard_id in $(seq -f "%03g" 0 1); do +for shard_id in $(seq -f "%03g" 0 13); do uv run data/generate_fw_edu_qa_v2.py --shard_pattern "${shard_id}_00000" --n_qa_pairs=5 --vllm_model=google/gemma-3-12b-it --max_length=2000 --max_model_length=2048; uv run data/generate_fw_edu_qa_v2_repeat.py --shard_pattern "min_0_to_2000/${shard_id}*level_0*" --n_qa_pairs=5 --vllm_model=google/gemma-3-12b-it; diff --git a/scripts/main_exp/gen_data_test.sh b/scripts/main_exp/gen_data_test.sh new file mode 100755 index 0000000..f2adb4f --- /dev/null +++ b/scripts/main_exp/gen_data_test.sh @@ -0,0 +1,20 @@ +# download fineweb_edu to `data/raw_datasets/fineweb_edu +uv run data/download_fineweb_edu.py + +# generate qa data +# run from 000 to 013 +for shard_id in $(seq -f "%03g" 0 1); do + uv run data/generate_fw_edu_qa_v2.py --shard_pattern "${shard_id}_00000" --n_qa_pairs=5 --vllm_model=google/gemma-3-12b-it --max_length=2000 --max_model_length=2048; + uv run data/generate_fw_edu_qa_v2_repeat.py --shard_pattern "min_0_to_2000/${shard_id}*level_0*" --n_qa_pairs=5 --vllm_model=google/gemma-3-12b-it; + + # self-generated response QA data + uv run data/self_generate_qa.py --vllm_model google/gemma-2-2b-it --glob_pattern "data/raw_datasets/fw_qa_v2/min_0_to_2000/${shard_id}*_level_1*" --closed_qa_prob 1.0 +done + + +# val split +uv run data/self_generate_qa.py --vllm_model google/gemma-2-2b-it --glob_pattern 'data/raw_datasets/fw_qa_v2/min_0_to_2000/*_level_0_val.parquet' + +# self-gen data for other ds +uv run data/self_generate_qa.py --vllm_model google/gemma-2-2b-it --ds_names squad_compact ropes_compact drop_compact --closed_qa_prob 1.0 +uv run data/self_generate_qa.py --vllm_model google/gemma-2-2b-it --ds_names pwc_compact --closed_qa_prob 0.0 diff --git a/scripts/main_exp/train.sh b/scripts/main_exp/train.sh old mode 100644 new mode 100755 diff --git a/scripts/niah/0-gen_data.sh b/scripts/niah/0-gen_data.sh old mode 100644 new mode 100755 diff --git a/scripts/niah/1-train.sh b/scripts/niah/1-train.sh old mode 100644 new mode 100755 diff --git a/scripts/niah/2-eval.sh b/scripts/niah/2-eval.sh old mode 100644 new mode 100755 diff --git a/src/ctx_to_lora/__init__.py b/src/ctx_to_lora/__init__.py index 115684e..e69de29 100644 --- a/src/ctx_to_lora/__init__.py +++ b/src/ctx_to_lora/__init__.py @@ -1 +0,0 @@ -from . import monkey_patch as monkey_patch