Instructions to use modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126")
model = AutoModelForCausalLM.from_pretrained("modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126

SGLang

How to use modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126 with Docker Model Runner:
```
docker model run hf.co/modelcollapse/gemma-3-270m-it-cpt-annealing-sft-251126
```

🧪 Model Behavior Incident Log

Collapse Symptoms After Annealing + SFT (Gemma-3-270M-TW)

This incident report documents multiple unexpected behaviors observed during the following pipeline:

Gemma-3-270M (base) → CPT → Dialogue-Annealing → SFT
Training performed with LLaMA-Factory on multiple Traditional Chinese datasets.

The purpose of this log is to document:

what happened
why the model collapsed
representative failure outputs
suspected root causes
mitigation steps and future plans

📉 Summary of the Incident

After the stage CPT → Phase-3 Dialogue Annealing → SFT, the model displayed severe conversational degradation, including:

repetitive loops
meaningless vocalization (e.g., hihihihi...)
empty or nearly empty responses
echoing the user query
hallucinated list expansions
loss of chat-template consistency
collapse-like distribution drift

These issues did not appear in earlier checkpoints (e.g., CPT → SFT without annealing).

🔍 Representative Failure Outputs

1. Character-level repetition

Prompt:

hi

Output:

hihihihihihihihihihihihihihihi...

2. Garbled symbols / noise

Prompt:

你好

Output:

總！！」！」！」！」!”！」！”！”！」!”」!”

3. Semantic drift + self-looping

Prompt:

圓的積分公式

Output:

是什麼？
有，他們都是那個。
你們都 是那個。
我 還是在那個裡頭，
所以，沒了，沒有，沒了。。
我也不知道。
我知道。
知道。
知道了。

4. Query echoing + hallucinated QA scaffolding

Prompt:

請用一句話解釋地球為什麼會自轉。

Output:

請用一句話解釋地球為什麼會自轉。
- 根據物理學，什麼是「氣體」？
- 什麼是「磁場」？

5. Baseline sanity check (pre-annealing) behaves normally

Prompt:

你好阿

Earlier output:

你好！很高興能和你一起交流，我是台灣大語言模型 Formosa-1...

🧭 Suspected Root Causes

1. Annealing Dataset Format Drift

Use of raw:

user: ...
assistant: ...

instead of structured JSON messages.

2. Overwriting Instruction Alignment

Annealing is fragile for 270M models → possible distribution collapse.

3. SFT Unable to Recover

Loss decreased, but entropy collapse persisted.

🛠 Actions Taken

Rebuilt CPT dataset (2.4M samples, 70/30 mix)
Adopted stable LR tail (cosine_with_min_lr)
Will rebuild annealing data with strict chat format
Full SFT pipeline will be repeated after clean CPT

🔄 Next Steps

Complete clean CPT
Rebuild dialogue annealing
Annealing from min_lr → 0
Re-run SFT
Compare ablation paths
Add regression tests

🧭 Closing Note

Small multilingual models are extremely sensitive to formatting + LR scheduling.
This report documents the collapse event for transparency and future reproducibility.