Text Classification
Transformers
Safetensors
English
llama
feature-extraction
llama3
reward-model
preference-modeling
rlhf
multi-domain
coherence
commonsense
empathy
multicultural
shared-prompt-gating
custom_code
text-embeddings-inference
Instructions to use mario-rc/multi-domain-rm-fsfairx-llama-3-8b-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use mario-rc/multi-domain-rm-fsfairx-llama-3-8b-it with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="mario-rc/multi-domain-rm-fsfairx-llama-3-8b-it", trust_remote_code=True)# Load model directly from transformers import AutoTokenizer, AutoModel tokenizer = AutoTokenizer.from_pretrained("mario-rc/multi-domain-rm-fsfairx-llama-3-8b-it", trust_remote_code=True) model = AutoModel.from_pretrained("mario-rc/multi-domain-rm-fsfairx-llama-3-8b-it", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Restore complete model card and metadata
Browse filesRestore the detailed description, dataset and task metadata, training-data documentation, evaluation section, and scoring Spearman results.
README.md
CHANGED
|
@@ -134,12 +134,12 @@ print({"chosen": chosen_score.item(), "rejected": rejected_score.item()})
|
|
| 134 |
For padded batches, also pass the matching `attention_mask`. Scores are intended for comparison
|
| 135 |
within a prompt; they are not calibrated probabilities or universal utility values.
|
| 136 |
|
| 137 |
-
##
|
| 138 |
|
| 139 |
-
This
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
|
| 144 |
## Credits
|
| 145 |
|
|
|
|
| 134 |
For padded batches, also pass the matching `attention_mask`. Scores are intended for comparison
|
| 135 |
within a prompt; they are not calibrated probabilities or universal utility values.
|
| 136 |
|
| 137 |
+
## Limitations
|
| 138 |
|
| 139 |
+
This is a reward model, not a standalone chat assistant. Scores are intended for relative comparison
|
| 140 |
+
and should be calibrated for each downstream use case. Performance can vary by language, topic, and
|
| 141 |
+
distribution. The model inherits limitations and biases from its base model and training data and
|
| 142 |
+
should not be used as the sole decision-maker in high-impact settings.
|
| 143 |
|
| 144 |
## Credits
|
| 145 |
|