File size: 1,469 Bytes
dc86e44 9248bfc dc86e44 a1e1c68 41724d3 dc86e44 7f69343 dc86e44 7f69343 dc86e44 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 |
---
model-index:
- name: gpt2-medium
results:
- task:
type: text-generation
dataset:
name: Wikitext
type: wikitext
metrics:
- type: perplexity (BASELINE)
value: 26.746555798139067
- type: perplexity (BASIC)
value: 30.20611307250951
---
This is a d-Matrix functional reference of the GPT2-MEDIUM model.
The reference provides the following functional *configurations*:
Configuration | Explanation
:-- | :--
**`BASELINE`** | a reference functionally equivalent to the original model
**`BASIC`** | all linear algebraic operands quantized to `MXINT8-64`, and all other operations transformed to approximated kernel simulations
### Usage
Install d-Matrix [Dmx_Compressor](https://github.com/d-matrix-ai/dmx-compressor) first.
```sh
pip install dmx_compressor
```
The following is an example model and its evaluation.
```sh
git clone https://github.com/EleutherAI/lm-evaluation-harness
cd lm-evaluation-harness
pip install -e .
```
```python
from dmx.compressor.modeling import DmxModel
import lm_eval
model_args = "pretrained='d-matrix/gpt2-medium',trust_remote_code=True"
lm = lm_eval.api.registry.get_model("hf").create_from_arg_string(model_args, {"batch_size": 1})
# Transform the model with DMX
lm._model = DmxModel.from_torch(lm._model).to_basic_model() # Using BASIC configuration
eval_results = lm_eval.evaluate(lm, lm_eval.tasks.get_task_dict([task])) # Assign desired task, i.e. "wikitext"
``` |