Update README.md

README.md

@@ -11,7 +11,7 @@ metrics:
 <!-- 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. -->
 
-# HC3-textgen-qa-pythia-6.9b-deduped-r1
+# pythia-6.9b-deduped for general QA
 
 This model is a fine-tuned version of [EleutherAI/pythia-6.9b-deduped](https://huggingface.co/EleutherAI/pythia-6.9b-deduped) on the pszemraj/HC3-textgen-qa dataset.
 It achieves the following results on the evaluation set:
@@ -22,10 +22,41 @@ It achieves the following results on the evaluation set:
 
 Text generation model trained on the HC3 text data of human questions + chatGPT answers.
 
+### Usage
+
+Install necessary packages for inference (_unless you have a big boi GPU_)
+```bash
+pip install -U -q transformers bitsandbytes accelerate
+```
+
+Basic inference example:
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("pszemraj/pythia-6.9b-HC3")
+
+model = AutoModelForCausalLM.from_pretrained(
+    "pszemraj/pythia-6.9b-HC3", load_in_8bit=True, device_map="auto"
+)
+
+prompt = "I was wondering how much wood a woodchuck could chuck? <answer>"
+inputs = tokenizer(prompt, return_tensors="pt")
+inputs = inputs.to("cuda")
+outputs = model.generate(**inputs, max_new_tokens=300)
+result = tokenizer.batch_decode(outputs, skip_special_tokens=True)
+
+import pprint as pp
+
+pp.pprint(result[0])
+```
+
+The defautl `GenerationConfig` uses contrastive search with `top_k=4` and `penalty_alpha=0.6`. For more information on inference and parameters to use, see [the transformers docs](https://huggingface.co/docs/transformers/generation_strategies#decoding-strategies).
 
 ## Intended uses & limitations
 
-More information needed
+- **Intended use:** research/exploration into comparing RLHF tuning vs. "guided"/specific tuning on "quality" datasets/responses of _"what the human would want as answer anyway"_
+- This is not trained/fine-tuned with RLHF and therefore will not be as helpful/generalizable/safe as chatGPT.
 
 ## Training and evaluation data