zai-org/Vision2Web
Viewer β’ Updated β’ 193 β’ 782 β’ 22
Hu Zang
zanghu
Β·
AI & ML interests
None yet
Recent Activity
liked a dataset 1 day ago
zai-org/Vision2Web liked a dataset 21 days ago
programbench/ProgramBench-Tests liked a dataset about 1 month ago
skylenage-ai/QwenClawBenchOrganizations
None yet
liked a
dataset 1 day ago
liked a
dataset 21 days ago
liked 2
datasets about 1 month ago
liked a
model over 1 year ago
liked a
dataset over 1 year ago
reacted to joaogante's post with π€π over 1 year ago
Post
4155
New sampling strategy dropped in π€ transformers -- Min P sampling π₯
Are you tired of having
Min P consists of a dynamic token filter -- as opposed to Top K, which keeps the K most likely tokens, and Top P, which keeps the most likely tokens up to a fixed cumulative probability, both static filters. Min P takes a base probability (defined in the
π High probability token present -> aggressive filter (we don't want to miss on that high-probability case and risk derailing generation)
π No high probability token present -> relaxed filter (there are many continuation possibilities that the model finds plausible)
You should set
Kudos to @kalomaze and @menhguin for creating this technique π₯ Read their discussion in the original issue for benchmarks (https://github.com/huggingface/transformers/issues/27670)
Copy-pasteable version of the example in the image below here: https://pastebin.com/VqXNtuxd
Have fun experimenting! π
Are you tired of having
top_k arbitrarily discarding high-quality continuations? Or top_p forgetting to exclude low-probability tokens, derailing your generation? Try out the new min_p flag in generate, fresh from a PR merged today! π₯¬Min P consists of a dynamic token filter -- as opposed to Top K, which keeps the K most likely tokens, and Top P, which keeps the most likely tokens up to a fixed cumulative probability, both static filters. Min P takes a base probability (defined in the
min_p flag) and multiplies it by the probability of the most likely token in the distribution for the next token. All tokens less likely than the resulting value are filtered. What happens with this strategy?π High probability token present -> aggressive filter (we don't want to miss on that high-probability case and risk derailing generation)
π No high probability token present -> relaxed filter (there are many continuation possibilities that the model finds plausible)
You should set
min_p to a low value, between 0.05 and 0.1. It behaves particularly well for creative text generation when paired up with temperature > 1.Kudos to @kalomaze and @menhguin for creating this technique π₯ Read their discussion in the original issue for benchmarks (https://github.com/huggingface/transformers/issues/27670)
Copy-pasteable version of the example in the image below here: https://pastebin.com/VqXNtuxd
Have fun experimenting! π
liked 2
Spaces over 1 year ago
FIM-Tokens not marked special
1
#4 opened over 1 year ago
by 
ruediste
liked a
model over 1 year ago
δΈΊδ»δΉθΏδΈͺ樑εηvacub.jsonζδ»Άι沑ζδΈζοΌηζ
4
#1 opened almost 4 years ago
by 
GordonTaoHuang
liked 2
datasets over 1 year ago
liked a
Space about 2 years ago
liked 3
models over 2 years ago
liked a
dataset over 2 years ago