Code

OpenELM support#7359

Code

Merged

ggerganov merged 16 commits intoggerganov:masterggerganov/llama.cpp:masterfrom icecream95:openelmicecream95/llama.cpp:openelmCopy head branch name to clipboardJul 4, 2024

+675-175

Commits

Commits on May 18, 2024

• Initial OpenELM support (270M only so far)

  icecream95

  committed
  217d8d7

• Fill out missing entries in llama_model_type_name

  icecream95

  committed
  aaabe2e

• fixup! Initial OpenELM support (270M only so far)

  Show description for 60b2e1b
  icecream95

  committed
  60b2e1b

Commits on Jun 30, 2024

• Merge branch 'master' into openelm

  compilade

  committed
  51b2577

Commits on Jul 1, 2024

• llama : support all OpenELM models

  Show description for c8cdb48
  compilade

  committed
  c8cdb48

• llama : minor spacing changes

  Show description for e3e33c0
  

  compilade

  and

  ggerganov

  authored
  e3e33c0

Commits on Jul 4, 2024

• llama : use std::array for per-layer hparams

  ggerganov

  committed
  29ab5a0

• llama : fix save/load state

  ggerganov

  committed
  b59ddf9

• llama : do not print hparams for vocab-only models

  ggerganov

  committed
  9971c38

• Merge branch 'master' into pr/7359

  ggerganov

  committed
  22a648f

• llama : handle n_head == 0

  ggerganov

  committed
  3fe395d

• Merge branch 'master' into pr/7359

  ggerganov

  committed
  199d0fb

• llama : use const ref for print_f and fix division by zero

  compilade

  committed
  269e07b

• Merge branch 'master' into openelm

  compilade

  committed
  c6ac198

• llama : fix t5 uses of n_head and n_ff

  compilade

  committed
  18e9287

• llama : minor comment

  ggerganov

  committed
  8be4fe4