|
|
--- |
|
|
license: apache-2.0 |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- text-generation-inference |
|
|
- transformers |
|
|
- smolify |
|
|
- dslm |
|
|
pipeline_tag: text-generation |
|
|
inference: |
|
|
parameters: |
|
|
temperature: 1 |
|
|
top_p: 0.95 |
|
|
top_k: 64 |
|
|
--- |
|
|
# 🤏 smolified-transcripts-checker |
|
|
|
|
|
> **Intelligence, Distilled.** |
|
|
|
|
|
This is a **Domain Specific Language Model (DSLM)** generated by the **Smolify Foundry**. |
|
|
|
|
|
It has been synthetically distilled from SOTA reasoning engines into a high-efficiency architecture, optimized for deployment on edge hardware (CPU/NPU) or low-VRAM environments. |
|
|
|
|
|
## 📦 Asset Details |
|
|
- **Origin:** Smolify Foundry (Job ID: `3a495556`) |
|
|
- **Architecture:** DSLM-Micro (270M Parameter Class) |
|
|
- **Training Method:** Proprietary Neural Distillation |
|
|
- **Optimization:** 4-bit Quantized / FP16 Mixed |
|
|
- **Dataset:** [Link to Dataset](https://huggingface.co/datasets/sugatobagchi/smolified-transcripts-checker) |
|
|
|
|
|
## 🚀 Usage (Inference) |
|
|
This model is compatible with standard inference backends like vLLM. |
|
|
|
|
|
```python |
|
|
# Example: Running your Sovereign Model |
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
|
|
model_id = "sugatobagchi/smolified-transcripts-checker" |
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id) |
|
|
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto") |
|
|
|
|
|
messages = [ |
|
|
{'role': 'system', 'content': '''Client Style Guide: 1. Numbers 0-9 spelled out, 10+ digits. 2. Speaker labels in all caps followed by a colon. 3. US currency uses '$' prefix. 4. Dates in 'Month Day, Year' format. 5. 'uh' and 'um' removed. 6. Sentences with any potential error, ambiguity, or awkwardness are flagged with '[??]' at the end.'''}, |
|
|
{'role': 'user', 'content': '''Interviewer: We're talking about four key points here. Participant: Yes. Exactly four points to cover. Not three, not five.'''} |
|
|
] |
|
|
text = tokenizer.apply_chat_template( |
|
|
messages, |
|
|
tokenize = False, |
|
|
add_generation_prompt = True, |
|
|
).removeprefix('<bos>') |
|
|
|
|
|
from transformers import TextStreamer |
|
|
_ = model.generate( |
|
|
**tokenizer(text, return_tensors = "pt").to("cuda"), |
|
|
max_new_tokens = 1000, |
|
|
temperature = 1, top_p = 0.95, top_k = 64, |
|
|
streamer = TextStreamer(tokenizer, skip_prompt = True), |
|
|
) |
|
|
``` |
|
|
|
|
|
## ⚖️ License & Ownership |
|
|
This model weights are a sovereign asset owned by **sugatobagchi**. |
|
|
Generated via [Smolify.ai](https://smolify.ai). |
|
|
|
|
|
[<img src="https://smolify.ai/smolify.gif" width="100"/>](https://smolify.ai) |
|
|
|
