From f07b38d5da0db0adf4d6920e86b951ba86c3b8d8 Mon Sep 17 00:00:00 2001
From: 51616 <rujikorn.ch@gmail.com>
Date: Fri, 27 Jun 2025 01:57:15 +0900
Subject: [PATCH] more configs

---
 configs/qa_short_ctx.yaml                     | 59 +++++++++++++++++++
 configs/qa_short_ctx_self_gen.yaml            | 52 ++++++++++++++++
 ...ml => qa_short_ctx_self_gen_no_fw_qa.yaml} |  0
 3 files changed, 111 insertions(+)
 create mode 100644 configs/qa_short_ctx.yaml
 create mode 100644 configs/qa_short_ctx_self_gen.yaml
 rename configs/{self_gen_qa_short_ctx_no_fw_qa.yaml => qa_short_ctx_self_gen_no_fw_qa.yaml} (100%)

diff --git a/configs/qa_short_ctx.yaml b/configs/qa_short_ctx.yaml
new file mode 100644
index 0000000..f57304b
--- /dev/null
+++ b/configs/qa_short_ctx.yaml
@@ -0,0 +1,59 @@
+output_dir: "" # just a placeholder
+bf16: true
+model_name_or_path: google/gemma-3-1b-it
+label_names: ["labels"]
+# eval_on_start: True
+# eval_strategy: "steps"
+# eval_steps: 500
+# save_strategy: "no"
+# # save_steps: 500
+# logging_strategy: "steps"
+# logging_steps: 100
+# use_liger_kernel: true
+# remove_unused_columns: false
+
+# needed to avoid OOM by compute the metrics batch by batch
+# w/o this the trainer stores logits of all sample in memory...
+# batch_eval_metrics: true
+
+per_device_train_batch_size: 8
+per_device_eval_batch_size: 8
+max_val_samples_per_ds: 1000
+# optim: schedule_free_adamw
+
+learning_rate: 0.00004
+# lr_scheduler_type: "constant_with_warmup"
+neftune_noise_alpha: 1
+weight_decay: 0.01
+
+warmup_steps: 100
+
+dataloader_prefetch_factor: 16
+dataloader_num_workers: 8
+
+# LoRA
+lora_r: 8
+lora_dropout: 0.0
+target_modules:
+  - down_proj
+
+# short ctx data
+train_ds_names:
+  - fw_qa_v2_2k_len_level_3
+  - squad_compact
+  - pwc_compact
+  - drop_compact
+  - ropes_compact
+
+# the contexts are kinda long-ish
+# - narrativeqa
+# - quoref
+# - synthetic_convqa 
+
+val_ds_names:
+  - fw_qa_v2_2k_len_level_0
+  - squad
+  - drop
+  - ropes
+  - pwc
+
diff --git a/configs/qa_short_ctx_self_gen.yaml b/configs/qa_short_ctx_self_gen.yaml
new file mode 100644
index 0000000..d8608ae
--- /dev/null
+++ b/configs/qa_short_ctx_self_gen.yaml
@@ -0,0 +1,52 @@
+output_dir: "" # just a placeholder
+bf16: true
+model_name_or_path: google/gemma-3-1b-it
+label_names: ["labels"]
+# eval_on_start: True
+# eval_strategy: "steps"
+# eval_steps: 500
+# save_strategy: "no"
+# # save_steps: 500
+# logging_strategy: "steps"
+# logging_steps: 100
+# use_liger_kernel: true
+# remove_unused_columns: false
+
+# needed to avoid OOM by compute the metrics batch by batch
+# w/o this the trainer stores logits of all sample in memory...
+# batch_eval_metrics: true
+
+per_device_train_batch_size: 8
+per_device_eval_batch_size: 8
+max_val_samples_per_ds: 1000
+# optim: schedule_free_adamw
+
+learning_rate: 0.00004
+# lr_scheduler_type: "constant_with_warmup"
+neftune_noise_alpha: 1
+weight_decay: 0.01
+
+warmup_steps: 100
+
+dataloader_prefetch_factor: 16
+dataloader_num_workers: 8
+
+# LoRA
+lora_r: 8
+lora_dropout: 0.0
+target_modules:
+  - down_proj
+
+# data
+train_ds_names:
+  - self_gen/gemma-2-2b-it/fw_qa_v2_2k_len_level_3
+  - self_gen/gemma-2-2b-it/squad_compact
+  - self_gen/gemma-2-2b-it/pwc_compact
+  - self_gen/gemma-2-2b-it/ropes_compact
+  - self_gen/gemma-2-2b-it/drop_compact
+
+val_ds_names:
+  - squad
+  - pwc
+  - drop
+  - ropes
diff --git a/configs/self_gen_qa_short_ctx_no_fw_qa.yaml b/configs/qa_short_ctx_self_gen_no_fw_qa.yaml
similarity index 100%
rename from configs/self_gen_qa_short_ctx_no_fw_qa.yaml
rename to configs/qa_short_ctx_self_gen_no_fw_qa.yaml