Update README.md

README.md

@@ -6,7 +6,7 @@ If you are new to the project, this document explains **where the data comes fro
 
 ## Provenance
 
-- **Base model:** `/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/不冻结Qwen训练/models/Qwen2.5-1.5B-Instruct`
+- **Base model:** `Qwen2.5-1.5B-Instruct`
 - **Datasets:** sampled from `../../prepare/data/math/*.json`. Each JSON is a list of `{prompt, response, system?}` records. `dataset_sampler.py` draws 10 disjoint groups of 100 samples (unless the dataset has <1 000 examples, in which case sampling with replacement keeps the group size fixed) using a deterministic seed derived from the dataset name.
 - **Training recipe (from `config/default.yaml`):**
   - sequence length 4 096; LoRA `r=64`, `alpha=128`, `dropout=0.05`, target modules = `{q,k,v,o,gate,up,down}_proj`
@@ -111,7 +111,7 @@ When inspecting or sharing a run, the **minimum** file set is `adapter/` + `prom
 If someone wants to regenerate any adapter from scratch:
 
 ```bash
-cd /hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora
+cd train_lora
 python -m train_lora.dataset_sampler --overwrite   # regenerates prompt groups
 python -m train_lora.train_single --dataset Math_QA --group 0
 # or run the full queue
@@ -127,7 +127,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 from peft import PeftModel
 import torch
 
-base_model = "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/不冻结Qwen训练/models/Qwen2.5-1.5B-Instruct"
+base_model = "Qwen2.5-1.5B-Instruct"
 adapter_dir = "outputs/Math_QA/group_00/adapter"
 
 tokenizer = AutoTokenizer.from_pretrained(adapter_dir, trust_remote_code=True)
@@ -172,7 +172,7 @@ trainer.train(resume_from_checkpoint="outputs/Math_QA/group_00/checkpoints/check
 The evaluation stack in `../评估体系` and `../parameter_generator/评估` expects this directory layout. Example:
 
 ```bash
-cd /hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/评估体系
+cd 评估体系
 python scripts/run_all_evals.py \
   --config configs/eval_config.yaml \
   --datasets Math_QA \