Upload merged Qwen3-4B-Instruct-2507 model (auto-generated README)

README.md

@@ -1,6 +1,7 @@
 ---
 base_model: Qwen/Qwen3-4B-Instruct-2507
 datasets:
+- u-10bei/dbbench_sft_dataset_react_v4
 - u-10bei/sft_alfworld_trajectory_dataset_v5
 language:
 - en
@@ -9,37 +10,67 @@ library_name: peft
 pipeline_tag: text-generation
 tags:
 - lora
+- peft
+- unsloth
 - agent
 - tool-use
+- agentbench
 - alfworld
 - dbbench
+- db-oversampling
 ---
 
-# qwen3-4b-agent-trajectory-lora
+# qwen3-4b-agentbench-dbalf-lora
 
-This repository provides a **LoRA adapter** fine-tuned from
-**Qwen/Qwen3-4B-Instruct-2507** using **LoRA + Unsloth**.
+This repository provides a **LoRA adapter** fine-tuned from **Qwen/Qwen3-4B-Instruct-2507**
+using **LoRA + Unsloth** for **AgentBench-style multi-turn agent trajectories**.
 
 This repository contains **LoRA adapter weights only**.
 The base model must be loaded separately.
 
 ## Training Objective
 
-This adapter is trained to improve **multi-turn agent task performance**
-on ALFWorld (household tasks) and DBBench (database operations).
+This adapter is trained to improve **multi-turn agent task performance** on:
 
-Loss is applied to **all assistant turns** in the multi-turn trajectory,
-enabling the model to learn environment observation, action selection,
-tool use, and recovery from errors.
+- **DBBench** (database operation / SQL generation trajectories)
+- **ALFWorld** (household task trajectories)
+
+Loss is applied to **all assistant turns** in the trajectory, enabling the model to learn:
+
+- environment observation
+- action selection
+- tool use / operation formatting
+- recovery from intermediate errors
+
+## Training Data
+
+- DBBench dataset: `u-10bei/dbbench_sft_dataset_react_v4`
+- ALFWorld dataset: `u-10bei/sft_alfworld_trajectory_dataset_v5`
+- Mixing ratio (pre-merge target): **DB:ALF = 1:1**
+
+### DB Oversampling (category-aware)
+Enabled: **True**
+
+DB category weights used during training-data preparation:
+
+- counting: 6
+- comparison: 4
+- ranking: 2
+- select: 1
+- insert: 1
+- update: 1
+- other: 1
 
 ## Training Configuration
 
 - Base model: Qwen/Qwen3-4B-Instruct-2507
 - Method: LoRA (full precision base)
 - Max sequence length: 2048
-- Epochs: 2
-- Learning rate: 2e-06
-- LoRA: r=64, alpha=128
+- Epochs: 1
+- Learning rate: 1e-06
+- LoRA: r=64, alpha=128, dropout=0.0
+- Per-device train batch size: 2
+- Gradient accumulation: 4
 
 ## Usage
 
@@ -60,9 +91,22 @@ model = AutoModelForCausalLM.from_pretrained(
 model = PeftModel.from_pretrained(model, adapter)
 ```
 
+## Notes
+
+- This repository is intended for **adapter-only** distribution.
+- Please ensure compliance with the **base model license/terms** in addition to this repository's license.
+- If you publish evaluation results, it is recommended to report:
+  - AgentBench task split / seeds
+  - DBBench / ALFWorld mix ratio
+  - DB oversampling settings
+  - decoding settings
+
 ## Sources & Terms (IMPORTANT)
 
-Training data: u-10bei/sft_alfworld_trajectory_dataset_v5
+Training data:
+- u-10bei/dbbench_sft_dataset_react_v4
+- u-10bei/sft_alfworld_trajectory_dataset_v5
 
-Dataset License: MIT License. This dataset is used and distributed under the terms of the MIT License.
-Compliance: Users must comply with the MIT license (including copyright notice) and the base model's original terms of use.
+Dataset license / terms:
+- Please follow the original license and terms of each dataset repository.
+- This adapter repository license (**apache-2.0**) applies to the adapter files in this repository.