Actions: huggingface/trl
Actions
Showing runs from all workflows
2,500+ workflow runs
2,500+ workflow runs
lm_head projection on tensor cores instead of in fp32
Build PR Documentation
#20142:
Pull request #6863
synchronize
by
qgallouedec
loss_type="cce", a fused linear cross-entropy loaded from the Hub
Build PR Documentation
#20141:
Pull request #6859
synchronize
by
qgallouedec
lm_head projection on tensor cores instead of in fp32
Tests (experimental)
#3746:
Pull request #6863
synchronize
by
qgallouedec
lm_head projection on tensor cores instead of in fp32
Secret Leaks
#12941:
Pull request #6863
synchronize
by
qgallouedec
loss_type="cce", a fused linear cross-entropy loaded from the Hub
Secret Leaks
#12940:
Pull request #6859
synchronize
by
qgallouedec
lm_head projection on tensor cores instead of in fp32
Tests
#20415:
Pull request #6863
synchronize
by
qgallouedec
loss_type="cce", a fused linear cross-entropy loaded from the Hub
Tests
#20414:
Pull request #6859
synchronize
by
qgallouedec