Update custom model files, README, and requirements

README.md

@@ -1,82 +1,67 @@
 ---
-library_name: transformers
-tags:
-- generated_from_trainer
-- trl
-- sft
+license: mit
+language:
+- en
 datasets:
-- generator
-model-index:
-- name: tiny-audio
-  results: []
+- speechbrain/LoquaciousSet
+base_model:
+- openai/whisper-large-v3-turbo
+- HuggingFaceTB/SmolLM3-3B
+pipeline_tag: automatic-speech-recognition
+tags:
+- asr
+- speech-recognition
+- audio
+- smollm
+- whisper
+- moe
 ---
 
-<!-- This model card has been generated automatically according to the information the Trainer had access to. You
-should probably proofread and complete it, then remove this comment. -->
+# Tiny Audio Model Card
+
+This model was born from a simple idea: what if anyone could train a powerful, modern speech recognition model for the price of a few coffees? This model is the result of the [Tiny Audio course](https://github.com/alexkroman/tiny-audio/blob/main/docs/course/0-course-overview.md), a free, hands-on guide to building your own ASR system from scratch.
+
+## The Story of this Model
 
-# tiny-audio
+This model isn't the product of a massive research lab with an unlimited budget. It's the result of a 24-hour training run on a single GPU, made possible by an efficient projector-only training approach. By combining the strengths of OpenAI's Whisper encoder (`openai/whisper-large-v3-turbo`) and a powerful language model (`HuggingFaceTB/SmolLM3-3B`), and only training a Mixture of Simple Adapters (MOSA) projector between them, we can create a high-quality ASR model with minimal resources.
 
-This model is a fine-tuned version of [](https://huggingface.co/) on the generator dataset.
-It achieves the following results on the evaluation set:
-- Loss: 0.4543
+This model is a testament to the power of open-source and the incredible tools and models that are now available to everyone.
 
-## Model description
+## Architecture
 
-More information needed
+```
+Audio (16kHz) → Whisper Encoder (frozen) → MoE Projector (trainable) → SmolLM3-3B (frozen) → Text
+```
 
-## Intended uses & limitations
+**MoE Projector (MOSA):**
+- Convolutional downsampling: 4x sequence compression via two stride-2 conv layers
+- Router: Linear→ReLU→Linear with dense softmax over 4 experts
+- Experts: 4 adapters, each Linear→ReLU→Linear (2048→4096→2048)
+- Output normalization: RMSNorm
 
-More information needed
+## Intended Use
 
-## Training and evaluation data
+This model is for you. It's for the curious, the builders, the learners. It's for anyone who wants to understand how modern AI works by getting their hands dirty. Use it to transcribe your podcasts, your meetings, your voice memos. But more importantly, use it as a starting point. Fork it, fine-tune it, break it, and make it your own.
 
-More information needed
+## Performance
 
-## Training procedure
+This model achieves a Word Error Rate (WER) of **12.14%** on the LoquaciousSet test set. It's not perfect, but it's a solid baseline that you can build on. See how it compares to other models on the [community leaderboard](https://github.com/alexkroman/tiny-audio#leaderboard).
 
-### Training hyperparameters
+## How to Use
 
-The following hyperparameters were used during training:
-- learning_rate: 0.0001
-- train_batch_size: 8
-- eval_batch_size: 32
-- seed: 123
-- gradient_accumulation_steps: 3
-- total_train_batch_size: 24
-- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
-- lr_scheduler_type: cosine
-- lr_scheduler_warmup_steps: 1000
-- training_steps: 20000
+```python
+from transformers import pipeline
 
-### Training results
+pipe = pipeline("automatic-speech-recognition", model="mazesmazes/tiny-audio", trust_remote_code=True)
 
-| Training Loss | Epoch | Step  | Validation Loss |
-|:-------------:|:-----:|:-----:|:---------------:|
-| 8.8546        | 0.05  | 1000  | 3.8075          |
-| 0.8286        | 0.1   | 2000  | 0.5193          |
-| 0.7909        | 0.15  | 3000  | 0.4701          |
-| 0.6955        | 0.2   | 4000  | 0.4581          |
-| 0.599         | 0.25  | 5000  | 0.4434          |
-| 0.6159        | 0.3   | 6000  | 0.4353          |
-| 0.5764        | 0.35  | 7000  | 0.4260          |
-| 0.602         | 0.05  | 8000  | 0.4298          |
-| 0.5363        | 0.1   | 9000  | 0.4430          |
-| 0.5643        | 0.15  | 10000 | 0.4636          |
-| 0.5135        | 0.2   | 11000 | 0.4423          |
-| 0.4419        | 0.25  | 12000 | 0.4473          |
-| 0.4848        | 0.3   | 13000 | 0.4539          |
-| 0.4692        | 0.35  | 14000 | 0.4481          |
-| 0.5154        | 0.05  | 15000 | 0.4482          |
-| 0.4736        | 0.1   | 16000 | 0.4522          |
-| 0.5097        | 0.15  | 17000 | 0.4537          |
-| 0.4729        | 0.2   | 18000 | 0.4542          |
-| 0.4142        | 0.25  | 19000 | 0.4543          |
-| 0.4718        | 0.3   | 20000 | 0.4543          |
+result = pipe("path/to/audio.wav")
+print(result["text"])
+```
 
+## How to Get Involved
 
-### Framework versions
+This project is more than just a model; it's a community. Here's how you can get involved:
 
-- Transformers 4.57.1
-- Pytorch 2.8.0+cu128
-- Datasets 4.4.1
-- Tokenizers 0.22.1
+- **Take the course**: The best way to start is to go through the [free 6-hour course](https://github.com/alexkroman/tiny-audio/blob/main/docs/course/0-course-overview.md) and train your own model.
+- **Share your results**: Add your model to the [leaderboard](https://github.com/alexkroman/tiny-audio#leaderboard) and share what you've learned.
+- **Join the conversation**: Ask questions, share your ideas, and connect with other builders in the [GitHub Discussions](https://github.com/alexkroman/tiny-audio/discussions).

handler.py

@@ -108,25 +108,7 @@ class EndpointHandler:
         if inputs is None:
             raise ValueError("Missing 'inputs' in request data")
 
+        # Pass through any parameters from request, let model config provide defaults
         params = data.get("parameters", {})
-        max_new_tokens = params.get("max_new_tokens", 128)
-        num_beams = params.get("num_beams", 1)
-        do_sample = params.get("do_sample", False)
-        length_penalty = params.get("length_penalty", 1.0)
-        repetition_penalty = params.get("repetition_penalty", 1.05)
-        no_repeat_ngram_size = params.get("no_repeat_ngram_size", 0)
-        early_stopping = params.get("early_stopping", True)
-        default_diversity = self.pipe.model.config.inference_diversity_penalty
-        diversity_penalty = params.get("diversity_penalty", default_diversity)
-
-        return self.pipe(
-            inputs,
-            max_new_tokens=max_new_tokens,
-            num_beams=num_beams,
-            do_sample=do_sample,
-            length_penalty=length_penalty,
-            repetition_penalty=repetition_penalty,
-            no_repeat_ngram_size=no_repeat_ngram_size,
-            early_stopping=early_stopping,
-            diversity_penalty=diversity_penalty,
-        )
+
+        return self.pipe(inputs, **params)