初始化项目，由ModelHub XC社区提供模型

Model: ermiaazarkhalili/Qwen3-4B-SFT-Fable5-Glint
Source: Original Platform

README.md

@@ -0,0 +1,94 @@
+---
+base_model:
+  - unsloth/Qwen3-4B
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+  - unsloth
+  - lora
+  - trl
+  - sft
+---
+
+# Qwen3-4B-SFT-Fable5-Glint
+
+A LoRA fine-tune of [`unsloth/Qwen3-4B`](https://huggingface.co/unsloth/Qwen3-4B), supervised fine-tuned on `ermiaazarkhalili/Fable-5-Glint-Clean` (private).
+
+| | |
+| --- | --- |
+| **Base model** | [`unsloth/Qwen3-4B`](https://huggingface.co/unsloth/Qwen3-4B) |
+| **Architecture** | `Qwen3ForCausalLM` |
+| **Parameters** | 4.0B |
+| **Training data** | `ermiaazarkhalili/Fable-5-Glint-Clean` (private) |
+| **Method** | LoRA supervised fine-tuning via [Unsloth](https://github.com/unslothai/unsloth) + [TRL](https://github.com/huggingface/trl) |
+
+## Usage
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "ermiaazarkhalili/Qwen3-4B-SFT-Fable5-Glint"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, dtype='auto', device_map='auto')
+
+messages = [{"role": "user", "content": "Explain gradient checkpointing in two sentences."}]
+inputs = tokenizer.apply_chat_template(
+    messages, add_generation_prompt=True, return_tensors='pt'
+).to(model.device)
+
+outputs = model.generate(inputs, max_new_tokens=256)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+## Training configuration
+
+| Setting | Value |
+| --- | --- |
+| LoRA rank (r) | 16 |
+| LoRA alpha | 16 |
+| Learning rate | 0.0002 |
+| Epochs | 3 |
+| Effective batch size | 8 (2 x 4 grad accum) |
+| Max sequence length | 4096 |
+| Base precision | 4-bit (QLoRA) |
+| Target modules | `q_proj`, `k_proj`, `v_proj`, `o_proj`, `gate_proj`, `up_proj`, `down_proj` |
+
+## Held-out evaluation
+
+Next-token accuracy on a deterministic held-out split of `ermiaazarkhalili/Fable-5-Glint-Clean` (n = 199 samples), scored against the base model [`unsloth/Qwen3-4B`](https://huggingface.co/unsloth/Qwen3-4B). Assistant tokens only; both models are scored identically.
+
+| Metric | Base | This model | Δ |
+| --- | ---: | ---: | ---: |
+| Top-1 accuracy | 0.5917 | 0.6910 | +0.0993 |
+| Top-5 accuracy | 0.8352 | 0.9125 | +0.0773 |
+
+A delta measures how far fine-tuning moved this model from its own starting point; it is not a ranking against other models, which start from different baselines.
+
+## Observed training loss
+
+Measured from our SLURM logs for this configuration. These are training-loss
+observations only — see the held-out evaluation above for measured accuracy.
+
+| SLURM job | Steps | First loss | Final loss |
+| --- | --- | --- | --- |
+| unlabelled | 1,554 | 1.4223 | 0.8867 |
+
+## Limitations
+
+- No benchmark evaluation has been run on this checkpoint. The only reported
+  numbers are training-loss observations.
+- Inherits the biases, knowledge cutoff and failure modes of the base model.
+- Fine-tuned on a single instruction-following dataset; behaviour outside that
+  distribution is untested.
+- LoRA adapters were merged into the base weights, so the merged model cannot
+  be detached from this fine-tune.
+
+## Reproducing
+
+Trained by `notebooks/fable_distillation_qwen3-4b_fable-glint_unsloth.ipynb`, executed non-interactively with
+papermill on a SLURM H100 partition (Unsloth + TRL, LoRA).
+
+---
+
+*Card generated from the training run's own configuration and logs by*
+*`scripts/generate_hub_model_card.py`.*