Update README with token match rate on text backbone

[sdeeptan-aws]

Description

Updated Qwen2.5-Omni-7B contrib model README with 100% token match accuracy on text backbone. Qwen2.5-Omni is a multimodal model supporting vision, audio, and text. AutoModelForCausalLM does not work for multimodal models — the specific text backbone class must be used to load the HF reference. Some multimodal configs may be missing attributes expected by the text backbone (e.g., output_attentions) and require config patching. With the correct text backbone extraction, the model achieves 100% token match.

Model Information

Model Name: Qwen2.5-Omni-7B
Model Architecture: Multimodal (omni — vision, audio, text) model (Qwen2-based decoder-only transformer text backbone)
Purpose: Multimodal understanding and text generation

Checklist

Required Components

• Accuracy Test (test/integration/test_model.py)
  • Validates model generation and coherence
  • Performance benchmarks (TTFT, throughput)
  • Test can compile and run the model on Neuron
• README.md with the following sections:
  • Usage Example: Clear code example showing how to use the model
  • Compatibility Matrix: Table showing tested Neuron SDK versions and instance types
  • Example Checkpoints: Links to compatible model checkpoints
  • Testing Instructions: Command to run the test suite for the model
• Source Code (src/)
  • Modeling code following NxD Inference patterns (unchanged in this PR)

Optional Components

• Unit Tests (CPU or Neuron-based)

Folder Structure

/contrib/models/Qwen2.5-Omni-7B/
  README.md
  /src
    modeling_qwen2_5_omni.py
  /test
    /integration
      test_model.py

Testing

Model was compiled and tested with TP=2, batch_size=1, seq_len=128, bfloat16. Text backbone validated only — vision/audio modalities not yet verified.

1. Text backbone extraction: AutoModelForCausalLM fails for multimodal models — must use the specific text backbone class
2. Config patching: Some multimodal configs are missing attributes expected by the text backbone (e.g., output_attentions) and need patching

Test Results:

Test	Status	Result
Smoke Test	✅ PASS	Model loads successfully
Token Matching	✅ PASS	100% match (text backbone)
TTFT (P50)	✅ PASS	50.15ms
Throughput	✅ PASS	19.82 tok/s

Compatibility

Tested with:

• Instance Type(s): Trn1
• Configuration: TP=2, batch_size=1, seq_len=128, bfloat16

Additional Information

• Omni-modal: Supports vision, audio, and text — text backbone validated independently
• AutoModelForCausalLM doesn't work: Multimodal models register with different auto classes. Use the specific text backbone class for HF reference loading.
• Config patching may be needed: Multimodal configs can be missing text-only attributes like output_attentions

Related Issues

N/A

vLLM Integration

• This model/feature is intended for use with vLLM
• Documentation includes vLLM registration instructions

---

By submitting this PR, I confirm that:

• I have read and followed the contributing guidelines
• This is a community contribution and may have limited testing compared to officially-supported models
• The code follows best practices and is well-documented
• All required components listed above are included

[aws-yishanm]

Approved because Readme and test were present.