CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-27 08:21:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : support Jamba hybrid Transformer-Mamba models (#7531)

Browse Source 
* wip: llama : separate recurrent states from the KV cache

This will be necessary to support Jamba
(and other recurrent models mixed with Attention).

Doesn't compile yet, and finding a slot isn't yet done correctly for recurrent states.

* llama : use std::find for seq_nodes in llama_rs_cache

* llama : state checkpoints for recurrent models

* llama : correctly handle more edge cases for the rs cache

* llama : rename many llama_kv_cache_* functions

* llama : remove useless return value for some llama_cache_* functions

* llama : rethink recurrent state cell counts

* llama : begin work on support for variable GQA

This will also be useful for Jamba if we consider the Mamba layers
to have 0 KV heads.

* llama : gracefully fail when not finding hybrid slot

* llama : support Jamba

* llama : fix BERT inference without KV cache

* convert-hf : check for unprocessed Jamba experts

* convert-hf : support Mini-Jamba conversion

* llama : fix Jamba quantization sanity checks

* llama : sequence-length-aware batch splitting

* llama : use equal-sequence-length sub-batches for recurrent models

* ggml : simplify SSM-related operators

* llama : make recurrent state slot allocation contiguous

* llama : adapt internal uses of batches to llama_ubatch

* llama : fix batch split output count for embeddings

* llama : minimize swaps when reordering logits

This reduces overhead when running hellaswag
on thousands of sequences with very small 100k params Mamba models.

* llama : fix edge case finding batch seq_id of split recurrent cell

This otherwise was a problem when running the HellaSwag benchmark
with small batch sizes, making it crash.

* llama : avoid copies for simple batch splits

* ggml : make ggml_ssm_scan not modify its source tensors

* llama : fix shared recurrent tail cell count for small ubatch sizes

Otherwise it was impossible to run the 'parallel' example with '-ub 1'
with a Mamba or Jamba model.

* llama : fix .base() compilation error on Windows

* llama : allow doing the equivalent of SSM_CONV with SUM_ROWS and MUL

* ggml : allow GGML_OP_CONCAT to work on non-contiguous tensors

The implementation already supported it,
and this makes Mamba's conv step slightly faster.

* mamba : fix non-contiguous usage of ggml_silu

* llama : session saving and reloading for hybrid models

* convert_hf : fix Jamba conversion

* llama : fix mixed signedness comparison

* llama : use unused n_embd_k_gqa in k_shift

This also slightly reduces the diff from the master branch

* llama : begin renaming llama_past back to llama_kv_cache

* llama : remove implicit recurrent state rollbacks

* llama : partially apply clang-format style

* convert : fix jamba conv1d shape squeezing

* graph : add back hybrid memory graph input

But this time it contains the sub-cache graph inputs.
This *should* make it easier to handle updating the inputs
when caching the graph (eventually).

* model : add Jamba to Mamba-specific hparams printing

* jamba : remove redundant nullptr initializations

* model : remove unnecessary prefix for tensor loading constants

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* model : use ggml_swiglu_split for Mamba

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* model : make falcon-h1 use shared mamba2 layer builder

* memory : avoid referring to KV in recurrent cache logs

* gguf-py : avoid adding duplicate tensor mappings for Jamba

Some of the tensor names are common with Llama4

---------

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
compilade
2025-07-09 14:59:57 -04:00
committed by GitHub
parent 98bab638fb
commit 4a5686da22
10 changed files with 622 additions and 423 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
gguf-py/gguf/constants.py
View File

@@ -330,6 +330,7 @@ class MODEL_ARCH(IntEnum):
ARWKV7 = auto()
MAMBA = auto()
MAMBA2 = auto()
JAMBA = auto()
XVERSE = auto()
COMMAND_R = auto()
COHERE2 = auto()
@@ -432,7 +433,10 @@ class MODEL_TENSOR(IntEnum):
SSM_CONV1D = auto()
SSM_X = auto()
SSM_DT = auto()
SSM_DT_NORM = auto()
SSM_A = auto()
SSM_B_NORM = auto()
SSM_C_NORM = auto()
SSM_D = auto()
SSM_NORM = auto()
SSM_OUT = auto()
@@ -635,6 +639,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.ARWKV7: "arwkv7",
MODEL_ARCH.MAMBA: "mamba",
MODEL_ARCH.MAMBA2: "mamba2",
MODEL_ARCH.JAMBA: "jamba",
MODEL_ARCH.XVERSE: "xverse",
MODEL_ARCH.COMMAND_R: "command-r",
MODEL_ARCH.COHERE2: "cohere2",
@@ -738,7 +743,10 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
MODEL_TENSOR.SSM_CONV1D: "blk.{bid}.ssm_conv1d",
MODEL_TENSOR.SSM_X: "blk.{bid}.ssm_x",
MODEL_TENSOR.SSM_DT: "blk.{bid}.ssm_dt",
MODEL_TENSOR.SSM_DT_NORM: "blk.{bid}.ssm_dt_norm",
MODEL_TENSOR.SSM_A: "blk.{bid}.ssm_a",
MODEL_TENSOR.SSM_B_NORM: "blk.{bid}.ssm_b_norm",
MODEL_TENSOR.SSM_C_NORM: "blk.{bid}.ssm_c_norm",
MODEL_TENSOR.SSM_D: "blk.{bid}.ssm_d",
MODEL_TENSOR.SSM_NORM: "blk.{bid}.ssm_norm",
MODEL_TENSOR.SSM_OUT: "blk.{bid}.ssm_out",
@@ -1738,6 +1746,34 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.SSM_NORM,
MODEL_TENSOR.SSM_OUT,
],
MODEL_ARCH.JAMBA: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.SSM_IN,
MODEL_TENSOR.SSM_CONV1D,
MODEL_TENSOR.SSM_X,
MODEL_TENSOR.SSM_DT,
MODEL_TENSOR.SSM_DT_NORM,
MODEL_TENSOR.SSM_A,
MODEL_TENSOR.SSM_B_NORM,
MODEL_TENSOR.SSM_C_NORM,
MODEL_TENSOR.SSM_D,
MODEL_TENSOR.SSM_OUT,
MODEL_TENSOR.FFN_GATE_INP,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_GATE,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.FFN_GATE_EXP,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
],
MODEL_ARCH.XVERSE: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,