When running the enrich_testcases method, it's possible to encounter a situation where the number of tokens exceeds the max_tokens limit.

ChatAFLndss / ChatAFL

Large Language Model guided Protocol Fuzzing (NDSS'24)

Apache License 2.0

275 stars 30 forks source link

When running the enrich_testcases method, it's possible to encounter a situation where the number of tokens exceeds the max_tokens limit. #6

Open wzk2239115 opened 6 months ago

wzk2239115 commented 6 months ago

Marti2203 commented 3 months ago

Hi, this can happen when the initial seed does not fit into the context window. Similarly, an incomplete test case can be generated due to the model's output window not being large enough. A solution is to change the models used, but this requires prompt engineering. Some initial experiments showed that this works and possibly we will release a branch with patches in the subsequent weeks.