github synced commits to refs/pull/802/merge at github/minimind from mirror
-
c47d6e1787 Merge 857bcbc4f2596b046e0f844ae381b54e88a3a374 into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to refs/pull/829/merge at github/minimind from mirror
-
ccfaee37da Merge 56d4cddfd8871ed1df043f27a5961bb662566ab1 into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to refs/pull/831/merge at github/minimind from mirror
-
fc5ca6c35b Merge 66aef88185ed50dc1458da2776b20db1f5a6e5ac into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to refs/pull/832/merge at github/minimind from mirror
-
941db51efe Merge 5b26153b75cf17786be449bfac986f3d261befdd into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to refs/pull/769/merge at github/minimind from mirror
-
2587feff95 Merge 182bff1bd86dfc118f25d077430b5ff212d2328a into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to refs/pull/833/merge at github/minimind from mirror
-
d8e66b034b Merge 56a21ed3c2311924c5a4f0aad50364da7e267d0e into cc312c1cc614bc371cd85dcbcbc1d3ba1590f364
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
- Compare 7 commits »
github synced commits to master at github/minimind-o from mirror
-
02ebfb33d2 [fix] top-1 router gradient
github synced commits to refs/pull/21/merge at github/minimind-o from mirror
-
79f7d56b7d Merge 03b593621b8a94c4698303d1be404877b65fee7c into 02ebfb33d25f8a1f3d8e25274b6e50ef9715fc23
-
02ebfb33d2 [fix] top-1 router gradient
- Compare 2 commits »
github synced commits to master at github/minimind-v from mirror
-
fa36e4861d [fix] top-1 router gradient
github
synced and deleted reference refs/tags/refs/pull/857/merge at github/minimind from mirror
github
synced and deleted reference refs/tags/refs/pull/859/merge at github/minimind from mirror
github synced commits to master at github/minimind from mirror
-
cc312c1cc6 [fix] unify Ctrl+C handling
-
e2ee878519 Merge pull request #857 from wbbeyourself/fix-ctrl-c-exception
-
b63a9565e3 [fix] restore top-1 router gradient #858
-
1aae25911b Merge pull request #859 from basil-k-aji-dev/fix/moe-router-gradient-topk1
-
1c876435ef [fix] keep the MoE router trainable at num_experts_per_tok=1
- Compare 6 commits »
github synced commits to refs/pull/762/merge at github/minimind from mirror
-
b962ea06ff Merge 743420daacc87b979e6db2f440e8a7f02733fc67 into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
-
21ec325dba [update] simplify safe math eval
- Compare 10 commits »
github synced commits to refs/pull/844/head at github/minimind from mirror
-
856a57cacb docs: document Apple Silicon MPS training
-
791b5150cc test: cover MPS device and training paths
-
7b554a516b feat: enable MPS across training entrypoints
-
618b72dc2a feat: add device-agnostic training utilities
-
7a9137d2e9 [update] trim verbose ddp comments
- Compare 13 commits »
github synced commits to refs/pull/844/merge at github/minimind from mirror
-
86472c9900 Merge 856a57cacb5483ea70d6c23273302e1ec90bc8b9 into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
856a57cacb docs: document Apple Silicon MPS training
-
791b5150cc test: cover MPS device and training paths
-
7b554a516b feat: enable MPS across training entrypoints
-
618b72dc2a feat: add device-agnostic training utilities
- Compare 13 commits »
github synced commits to refs/pull/708/merge at github/minimind from mirror
-
937e4174a4 Merge 138c367ddb2ae269ad4c1d81c1b0c62f9ffc5e8e into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
- Compare 4 commits »
github synced commits to refs/pull/769/merge at github/minimind from mirror
-
6bb19aedbc Merge 182bff1bd86dfc118f25d077430b5ff212d2328a into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
-
21ec325dba [update] simplify safe math eval
- Compare 10 commits »
github synced commits to refs/pull/851/merge at github/minimind from mirror
-
bbbfee7ab6 Merge 59658fb60c384e6cae18a8c9601d43e47af4f5bb into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
-
21ec325dba [update] simplify safe math eval
- Compare 5 commits »
github synced commits to refs/pull/852/merge at github/minimind from mirror
-
3681271736 Merge 655d4705ebec80d7a78a848bf0a35a9768aa45be into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
-
21ec325dba [update] simplify safe math eval
- Compare 7 commits »
github synced commits to refs/pull/831/merge at github/minimind from mirror
-
a458e70363 Merge 66aef88185ed50dc1458da2776b20db1f5a6e5ac into 7a9137d2e90294df80ce9178b89e82657e19f5a7
-
7a9137d2e9 [update] trim verbose ddp comments
-
7988bc81a3 Merge pull request #855 from basil-k-aji-dev/perf/ddp-followups
-
37ac206e9a [perf] skip the per-forward RoPE buffer broadcast in the RL trainers
-
21ec325dba [update] simplify safe math eval
- Compare 9 commits »