Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -177,6 +177,22 @@ model = AutoModelForCausalLM.from_pretrained(BASE, torch_dtype="bfloat16",
|
|
| 177 |
model = PeftModel.from_pretrained(model, "Jainamshahhh/chronocalc-4b").eval()
|
| 178 |
```
|
| 179 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 180 |
## Reproducibility and license
|
| 181 |
|
| 182 |
The corpus regenerates byte for byte from a single seed, because a row is a pure function of its
|
|
|
|
| 177 |
model = PeftModel.from_pretrained(model, "Jainamshahhh/chronocalc-4b").eval()
|
| 178 |
```
|
| 179 |
|
| 180 |
+
|
| 181 |
+
## Try it, and everything that backs it
|
| 182 |
+
|
| 183 |
+
**Live side by side demo:** https://huggingface.co/spaces/Jainamshahhh/chronocalc-demo
|
| 184 |
+
Enter your own input and watch the base model and ChronoCalc-4B answer it under identical greedy
|
| 185 |
+
decoding. The GPU backend scales to zero, so a cold first request takes about a minute.
|
| 186 |
+
|
| 187 |
+
| | Hugging Face | Kaggle |
|
| 188 |
+
|---|---|---|
|
| 189 |
+
| model | https://huggingface.co/Jainamshahhh/chronocalc-4b | https://www.kaggle.com/datasets/jainamshahhh/chronocalc-4b |
|
| 190 |
+
| dataset | https://huggingface.co/datasets/Jainamshahhh/chronocalc | https://www.kaggle.com/datasets/jainamshahhh/chronocalc-date-time-arithmetic |
|
| 191 |
+
|
| 192 |
+
Released on both platforms, with the scorer, every eval slice, the per row verdicts and
|
| 193 |
+
`significance.py` alongside, so every number on this page can be recomputed rather than
|
| 194 |
+
trusted.
|
| 195 |
+
|
| 196 |
## Reproducibility and license
|
| 197 |
|
| 198 |
The corpus regenerates byte for byte from a single seed, because a row is a pure function of its
|