kingbri 586bfc6744 ExllamaV2: Bump version ...

v0.3.2

Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>

2025-07-13 18:06:03 -04:00

..

conversion

Add basic support for Qwen3MoE

2025-05-01 20:23:33 +02:00

exllamav2_ext

Optimize paged cache defrag

2025-05-27 01:53:37 +02:00

experimental

Add some code for evaluating FPx (not enabled)

2024-09-28 16:07:39 +02:00

generator

Optimize paged cache defrag

2025-05-27 01:53:37 +02:00

hadamard

Add hadamard functions

2024-04-13 14:24:48 +02:00

server

Don't cache encoding in lefttrim_token

2024-06-24 03:14:54 +02:00

tokenizer

Fail if tokenizer.json not found

2025-05-27 17:10:35 +02:00

vlm

Fix rounding error in Pixtral image preprocessor

2025-06-05 01:32:36 +02:00

__init__.py

Pixtral: Add vision tower and preprocessor

2024-11-10 11:15:06 +01:00

architecture.py

Add basic support for Qwen3MoE

2025-05-01 20:23:33 +02:00

attn_params.py

Fix alt pos embeddings and block diagonal mask when flash-attn is disabled

2025-02-13 22:13:48 +01:00

attn.py

Fix Gemma3 head norm (RMS)

2025-04-11 00:18:06 +02:00

cache.py

Remove (experimental) Q-cache calibration feature

2024-09-08 15:13:00 +02:00

compat.py

More stream gymnastics

2024-09-23 17:28:55 +02:00

config.py

Add basic support for Qwen3MoE

2025-05-01 20:23:33 +02:00

conv.py

Support Conv3d

2024-11-16 11:26:40 +01:00

device.py

Gemma3 local RoPE fixes

2025-04-10 22:16:08 +02:00

embedding.py

Upcast MM embeddings when residual is FP32

2025-03-15 00:06:56 +01:00

ext.py

More stream gymnastics

2024-09-23 17:28:55 +02:00

headnorm.py

Fix Gemma3 head norm (RMS)

2025-04-11 00:18:06 +02:00

layernorm.py

Refactor architecture logic for code reuse between LLM/VLM

2024-11-03 22:34:25 +01:00

linear.py

Allow loading transposed unquantized linear layer

2025-03-15 00:06:56 +01:00

lora.py

Refactor architecture logic for code reuse between LLM/VLM

2024-11-03 22:34:25 +01:00

mlp.py

Add support for Mistral 3.1 VLM

2025-04-18 22:47:47 +02:00

model_init.py

Check length of gpu_split in model_init

2025-01-09 11:36:25 +01:00

model.py

Gemma3 local RoPE fixes

2025-04-24 00:31:40 +02:00

module.py

Refactor architecture logic for code reuse between LLM/VLM

2024-11-03 22:34:25 +01:00

moe_mlp.py

Add basic support for Qwen3MoE

2025-05-01 20:23:33 +02:00

mrope.py

Allow per-layer RoPE theta

2025-03-15 00:06:56 +01:00

parallel_decoder.py

Allow per-layer RoPE theta

2025-03-15 00:06:56 +01:00

pos_embedding.py

Allow component models to use learned pos embeddings without regarding LLM max_seq_len

2025-03-15 00:06:56 +01:00

py.typed

improve installation experience (#666)

2024-11-02 21:11:14 +01:00

rmsnorm.py

Refactor architecture logic for code reuse between LLM/VLM

2024-11-03 22:34:25 +01:00

rope.py

Allow per-layer RoPE theta

2025-03-15 00:06:56 +01:00

stloader.py

Fix loading in new thread

2024-09-18 19:46:56 +02:00

tensor_p.py

concatenate the sin and cos tensors (#758)

2025-04-11 22:11:13 +02:00

util.py

Add some debug functions

2024-10-02 23:58:33 +02:00

version.py

ExllamaV2: Bump version

2025-07-13 18:06:03 -04:00