Initial anonymous release: Qwen3.5-9B judge LoRA (subq+human, ckpt-294)

README.md

@@ -0,0 +1,70 @@
+---
+base_model: Qwen/Qwen3.5-9B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+  - lora
+  - peft
+  - judge
+  - video-evaluation
+  - anonymous-release
+---
+
+# Anonymous Release — Judge LoRA Adapter (Qwen3.5-9B)
+
+A LoRA adapter for **Qwen/Qwen3.5-9B** trained as a judge model that scores
+generated videos against physical-law sub-rubrics derived from text prompts.
+Released anonymously alongside the companion dataset
+[`anonymouscla/physground`](https://huggingface.co/datasets/anonymouscla/physground).
+
+## Files
+
+```
+adapter_config.json            # PEFT/LoRA config (base_model = Qwen/Qwen3.5-9B)
+adapter_model.safetensors      # LoRA weights (~167 MB)
+additional_config.json         # ms-swift extras (lora_dtype / lr ratios)
+training_args.json             # sanitized training hyperparameters
+```
+
+## Training
+
+| Item | Value |
+| --- | --- |
+| Base model | `Qwen/Qwen3.5-9B` |
+| Tuning method | LoRA via PEFT (rank 32, α 64, dropout 0.05) |
+| Target modules | All linear layers in the language model (vision tower frozen; merger limited to `linear_fc1`/`linear_fc2`) |
+| Precision | bf16 with gradient checkpointing |
+| Optimizer | AdamW (fused), lr = 1e-4, cosine schedule, warmup 5% |
+| Batch | 1 × 8 grad-accum × 4 GPUs (global batch 32) |
+| Epochs / steps | 1.0 epoch / 294 steps |
+| Best eval loss | 0.1063 (step 294) |
+| Framework | ms-swift 4.1.2, PEFT 0.19.1, DeepSpeed ZeRO-2 |
+
+The training mixture combines automatically derived sub-question judgements
+with human-rated samples (the `subq+human` split). See the companion
+anonymous dataset for prompts, physical-law tags, and example videos.
+
+## Usage
+
+```python
+from peft import PeftModel
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+base_id = "Qwen/Qwen3.5-9B"
+adapter_dir = "."  # this directory
+
+tokenizer = AutoTokenizer.from_pretrained(base_id)
+base = AutoModelForCausalLM.from_pretrained(base_id, torch_dtype="bfloat16", device_map="auto")
+model = PeftModel.from_pretrained(base, adapter_dir)
+model.eval()
+```
+
+The adapter expects the standard Qwen 3.5 chat template (`qwen3_5`) and a
+prompt that asks the judge to answer one or more sub-rubric questions about a
+candidate video frame/caption. Greedy decoding (`temperature = 0`) with
+`max_new_tokens = 64` matches the training-time generation config.
+
+## License
+
+The base model is released by its respective authors; this LoRA adapter is
+shared for anonymous review purposes. No identifying metadata is included.

adapter_config.json

@@ -0,0 +1,40 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-9B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [],
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": "^(model\\.language_model(?=\\.).*\\.(o_proj|out_proj|in_proj_qkv|gate_proj|k_proj|in_proj_z|down_proj|v_proj|q_proj|in_proj_b|up_proj|in_proj_a)|model\\.visual\\.merger(?=\\.).*\\.(linear_fc2|linear_fc1))$",
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:880543b4bbc572e58980d58690b23ea9262d7ecbdd980bdf5a9139dfe022c881
+size 174336432

additional_config.json

@@ -0,0 +1 @@
+{"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}

training_args.json

@@ -0,0 +1,53 @@
+{
+  "_comment": "Sanitized excerpt of the training configuration. Local paths and tracking IDs removed.",
+  "base_model": "Qwen/Qwen3.5-9B",
+  "model_type": "qwen3_5",
+  "template": "qwen3_5",
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "max_length": 8192,
+  "max_new_tokens": 64,
+  "tuner": {
+    "type": "lora",
+    "lora_rank": 32,
+    "lora_alpha": 64,
+    "lora_dropout": 0.05,
+    "lora_bias": "none",
+    "target_modules": "all-linear (language model only; vision merger limited to linear_fc1/linear_fc2)",
+    "use_dora": false,
+    "use_rslora": false,
+    "freeze_vit": true,
+    "freeze_aligner": false
+  },
+  "optimizer": {
+    "name": "adamw_torch_fused",
+    "learning_rate": 1e-4,
+    "weight_decay": 0.1,
+    "adam_beta1": 0.9,
+    "adam_beta2": 0.95,
+    "adam_epsilon": 1e-8,
+    "max_grad_norm": 1.0,
+    "lr_scheduler_type": "cosine",
+    "warmup_ratio": 0.05,
+    "aligner_lr": 2e-6
+  },
+  "training": {
+    "num_train_epochs": 1.0,
+    "per_device_train_batch_size": 1,
+    "gradient_accumulation_steps": 8,
+    "world_size": 4,
+    "global_batch_size": 32,
+    "bf16": true,
+    "gradient_checkpointing": true,
+    "seed": 42,
+    "data_seed": 42,
+    "deepspeed_zero_stage": 2,
+    "total_steps": 294,
+    "best_eval_loss": 0.1063,
+    "best_step": 294
+  },
+  "framework": {
+    "ms_swift_version": "4.1.2",
+    "peft_version": "0.19.1"
+  }
+}