初始化项目，由ModelHub XC社区提供模型

Model: srk0102200/AnimTOON-3B Source: Original Platform
2026-08-15 02:23:16 +08:00
commit d7ea36943e
8 changed files with 387 additions and 0 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,36 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
--- a/README.md
+++ b/README.md
@@ -0,0 +1,179 @@
+---
+language:
+- en
+license: mit
+library_name: transformers
+base_model: Qwen/Qwen2.5-3B-Instruct
+tags:
+- animation
+- lottie
+- svg
+- animtoon
+- vector-animation
+- text-to-animation
+- conversational
+- text-generation-inference
+datasets:
+- OmniLottie/MMLottie-2M
+pipeline_tag: text-generation
+---
+
+# AnimTOON-3B (v3): Token-Efficient Vector Animation Generation
+
+**3-4x fewer tokens than OmniLottie (CVPR 2026) for generating Lottie animations. Now with character animation support.**
+
+| | AnimTOON | OmniLottie |
+|---|---|---|
+| **Tokens (simple)** | **166** | 616 |
+| **Tokens (complex)** | **597** | 4095 |
+| **VRAM** | **5GB** | 15.2GB |
+| **FPS** | **30** | 8 |
+| **Model Size** | **3B LoRA** | 4B full |
+| **Custom Tokenizer** | **No** | Yes (40k tokens) |
+| **Accepts SVG** | **Yes** | No |
+
+## What is AnimTOON?
+
+AnimTOON is a compact, plain-text animation format that any LLM can generate. Instead of outputting 18,000+ tokens of raw Lottie JSON, AnimTOON describes animations in ~166-597 tokens of human-readable text.
+
+```
+anim fr=30 dur=120
+
+layer Logo shape
+  fill #000000
+  path sh x2
+  pos [0.5,0.5]
+  rot 0.0->-67 0.04->46 0.14->-31 0.28->0 ease=bounce
+  scale 0.0->[0,0] 0.14->[90,90] 0.28->[100,100] ease=smooth
+  opacity 0.0->0 0.14->100 ease=fade
+```
+
+This produces a complete animated .lottie file with bounce entrance, rotation wobble, and fade-in.
+
+## How to Use
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+tokenizer = AutoTokenizer.from_pretrained("srk0102200/AnimTOON-3B")
+model = AutoModelForCausalLM.from_pretrained(
+    "srk0102200/AnimTOON-3B",
+    dtype=torch.float16,
+    device_map="cuda"
+)
+
+prompt = "a red circle pulsing in the center with a smooth bounce"
+messages = [{"role": "user", "content": f"Generate AnimTOON animation: {prompt}"}]
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tokenizer(text, return_tensors="pt").to("cuda")
+
+with torch.no_grad():
+    out = model.generate(**inputs, max_new_tokens=512, temperature=0.7, do_sample=True)
+result = tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
+print(result)
+```
+
+## Convert to .lottie
+
+```python
+# Clone: git clone https://github.com/srk0102/AnimTOON.git
+import sys; sys.path.insert(0, 'src')
+from toon_animator import animtoon_to_dotlottie_full
+
+animtoon_to_dotlottie_full(result, "output.lottie")
+# Preview at https://lottiefiles.com/preview
+```
+
+## Animate Any SVG
+
+```python
+from lottie import parsers  # pip install lottie
+
+# Convert SVG to Lottie (perfect paths)
+anim = parsers.svg.parse_svg_file("your_logo.svg")
+lottie_dict = anim.to_dict()
+
+# Generate AnimTOON animations with the model
+# Apply animations to the Lottie layers
+# Output: .lottie file with real SVG shapes + AI animations
+```
+
+See full pipeline: [test_svg_pipeline.py](https://github.com/srk0102/AnimTOON/blob/master/test_svg_pipeline.py)
+
+## Benchmark Results (Measured)
+
+**Same prompt, same hardware:**
+
+| Test | AnimTOON Tokens | OmniLottie Tokens | Ratio |
+|------|----------------|-------------------|-------|
+| Apple logo bounce | 207 (41 shape + 166 anim) | 1113 | 5.4x fewer |
+| Smiley face complex | 597 | 4095 | 6.9x fewer |
+| Simple ball bounce | 176 | 616 | 3.5x fewer |
+
+**Dataset statistics (99,650 samples):**
+- Average raw Lottie JSON: 18,202 tokens
+- Average AnimTOON: 222 tokens
+- Token reduction: 98.8%
+
+## Current Status (v3)
+
+**v3 adds character animation support** trained on Spine + DragonBones skeletal data.
+
+The model now works for:
+- Icon/logo animations (pulse, bounce, spin, fade, wobble)
+- **Character idle/walk cycles (14 layers, coordinated)**
+- **Multi-part SVG animation (47-part crab demo)**
+- Correct color matching from text descriptions
+- SVG + animation pipeline with per-part anchor points
+
+**Limitations:**
+- No shape generation (requires SVG input)
+- Model output varies between runs (temperature-dependent)
+- Position animation on shape groups not yet supported
+- Not yet trained on facial expressions
+
+## Training Details
+
+| Parameter | Value |
+|-----------|-------|
+| Base Model | Qwen/Qwen2.5-3B-Instruct |
+| Method | LoRA (r=16, alpha=32) merged into base |
+| Version | v3 (final 3B Lite release) |
+| Training Data | 99,650 (MMLottie-2M) + 10,000 (layer-aware) + 984 (Spine/DragonBones) |
+| Hardware | 1x NVIDIA RTX 5060 Ti (16GB) |
+| Framework | Unsloth |
+| Token Reduction | 98.8% vs raw Lottie JSON |
+
+## Architecture: Why Animation-Only is Better
+
+> "Asking one model to draw AND animate is like asking one person to paint AND dance at the same time."
+
+AnimTOON separates concerns:
+- **SVG provides shapes** (perfect, no hallucination, 0 tokens)
+- **Model generates animation** (focused, token-efficient)
+- **Converter merges them** (deterministic, 100% valid output)
+
+OmniLottie generates everything in one model → hallucinated shapes, token bloat (2001 tokens for a "crab" that looks like binoculars).
+
+## Links
+
+- **GitHub:** [github.com/srk0102/AnimTOON](https://github.com/srk0102/AnimTOON)
+- **PitchHut:** [pitchhut.com/project/animtoon-lottie-animation](https://www.pitchhut.com/project/animtoon-lottie-animation)
+- **OmniLottie (comparison):** [arxiv.org/abs/2603.02138](https://arxiv.org/abs/2603.02138)
+- **MMLottie-2M Dataset:** [huggingface.co/datasets/OmniLottie/MMLottie-2M](https://huggingface.co/datasets/OmniLottie/MMLottie-2M)
+
+## Citation
+
+```bibtex
+@misc{sivaramakrishna2026animtoon,
+  title={AnimTOON: Token-Efficient Vector Animation Generation via Compact Text Format},
+  author={Siva RamaKrishna},
+  year={2026},
+  url={https://github.com/srk0102/AnimTOON}
+}
+```
+
+## License
+
+MIT License - see [LICENSE](https://github.com/srk0102/AnimTOON/blob/master/LICENSE)
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
--- a/config.json
+++ b/config.json
@@ -0,0 +1,69 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "float16",
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 70,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 2,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.3.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}
--- a/generation_config.json
+++ b/generation_config.json
@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "5.3.0"
+}
--- a/model.safetensors
+++ b/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e021ad2795299db080811ec567dbec9605db16c21a86a28b0cac9ea4cbdf53e
+size 6171926680
--- a/tokenizer.json
+++ b/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": true,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}