Text Generation
Transformers
PyTorch
English
hf_olmo
conversational
custom_code
natolambert commited on
Commit
d79d388
1 Parent(s): 3bcd9ac

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -4
README.md CHANGED
@@ -156,11 +156,15 @@ Compared to Tulu 2, DPO hyperparameters are the same. SFT is lower LR and 3 epoc
156
 
157
  ## Bias, Risks, and Limitations
158
 
159
- The adapted OLMo models do not include a specific safety filter or safety training data.
160
- While our model shows good scores relative to its peers on ToxiGen, like any base language model or fine-tuned model without safety filtering, it is relatively easy for a user to prompt these models to generate harmful and generally sensitive content.
161
- Such content can also be produced unintentionally, especially in the case of bias, so we recommend users consider the risks of applications of this technology.
 
 
 
 
 
162
 
163
- Otherwise, many facts from OLMo or any LLM will often not be true, so they should be checked.
164
 
165
 
166
  ## Citation
 
156
 
157
  ## Bias, Risks, and Limitations
158
 
159
+ This adapted OLMo model is a research artifact, not a consumer product.
160
+ It is intended to benefit the research community interested in understanding the safety properties of LLMs and developers building safety tools for LLMs.
161
+ For this reason, the model does not include a specific safety filter or safety training data.
162
+ While our model scores well relative to its peers on ToxiGen, it is possible for the model to generate harmful and sensitive content from some user prompts.
163
+ We recommend developers exercise caution and consider the risks of the applications of this technology.
164
+ Furthermore, developers should consider implementing safeguards for biases, privacy, and other potential harms when appropriate.
165
+ Finally, as with every LLM, OLMo may produce factual-sounding outputs that may not be true, so developers and users are encouraged to confirm such outputs before relying on them.
166
+ All users of this model are responsible for how they use the model.
167
 
 
168
 
169
 
170
  ## Citation