dlasher/Pocket-Darwin-180B:UD-Q4_K

dlasher/ Pocket-Darwin-180B:UD-Q4_K_XL

3 Downloads Updated 5 days ago

Requant to IQ4_NL, with combined_en_huge as prompt (English + Coding focus)

Pocket-Darwin-180B:UD-Q4_K_XL ... /

model

8dd7789f02ff · 11MB

Metadata

general.architecture

qwen4exp

qwen4exp
general.file_type

Q4_K_M

Q4_K_M
qwen4exp.attention.compress_ratios

[0, 0, 0, 4, 0, ...]

[0, 0, 0, 4, 0, ...]
qwen4exp.attention.head_count

24

24
qwen4exp.attention.head_count_kv

2

2
qwen4exp.attention.indexer.head_count

4

4
qwen4exp.attention.indexer.key_length

128

128
qwen4exp.attention.indexer.top_k

2048

2048
qwen4exp.attention.key_length

256

256
qwen4exp.attention.layer_norm_rms_epsilon

1e-06

1e-06
qwen4exp.attention.value_length

256

256
qwen4exp.block_count

48

48
qwen4exp.context_length

262144

262144
qwen4exp.embedding_length

2560

2560
qwen4exp.embedding_length_per_layer_input

160

160
qwen4exp.expert_count

512

512
qwen4exp.expert_feed_forward_length

640

640
qwen4exp.expert_shared_feed_forward_length

640

640
qwen4exp.expert_used_count

10

10
qwen4exp.full_attention_interval

4

4
qwen4exp.hyper_connection.count

4

4
qwen4exp.hyper_connection.low_rank

320

320
qwen4exp.ple.conv_kernel

4

4
qwen4exp.ple.eos_token_id

248044

248044
qwen4exp.ple.head_offsets

[0, 20000003, 40000026, 60000059, 80000106, ...]

[0, 20000003, 40000026, 60000059, 80000106, ...]
qwen4exp.ple.head_vocab_sizes

[20000003, 20000023, 20000033, 20000047, 20000059, ...]

[20000003, 20000023, 20000033, 20000047, 20000059, ...]
qwen4exp.ple.heads_per_ngram

8

8
qwen4exp.ple.image_token_id

248056

248056
qwen4exp.ple.layer_multipliers

[23703573157769, 20109073645365, 8052911324071]

[23703573157769, 20109073645365, 8052911324071]
qwen4exp.ple.layers

[1]

[1]
qwen4exp.ple.ngram_size

3

3
qwen4exp.rope.dimension_count

64

64
qwen4exp.rope.dimension_sections

[11, 11, 10, 0]

[11, 11, 10, 0]
qwen4exp.rope.freq_base

1e+07

1e+07
qwen4exp.ssm.conv_kernel

4

4
qwen4exp.ssm.group_count

16

16
qwen4exp.ssm.inner_size

6144

6144
qwen4exp.ssm.state_size

128

128
qwen4exp.ssm.time_step_rank

48

48
tokenizer.ggml.add_bos_token

false

false
tokenizer.ggml.bos_token_id

248044

248044
tokenizer.ggml.eos_token_id

248046

248046
tokenizer.ggml.merges

[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]

[Ġ Ġ, ĠĠ ĠĠ, i n, Ġ t, ĠĠĠĠ ĠĠĠĠ, ...]
tokenizer.ggml.model

gpt2

gpt2
tokenizer.ggml.padding_token_id

248044

248044
tokenizer.ggml.pre

qwen35

qwen35
tokenizer.ggml.token_type

[1, 1, 1, 1, 1, ...]

[1, 1, 1, 1, 1, ...]
tokenizer.ggml.tokens

[!, ", #, $, %, ...]

[!, ", #, $, %, ...]
quantize.imatrix.chunks_count

45

45
quantize.imatrix.dataset

unsloth_calibration_Qwen3.8-Flash-Next.txt

unsloth_calibration_Qwen3.8-Flash-Next.txt
quantize.imatrix.entries_count

926

926
quantize.imatrix.file

Qwen3.8-Flash-Next-GGUF/imatrix_unsloth.gguf

Qwen3.8-Flash-Next-GGUF/imatrix_unsloth.gguf
split.count

4

4
split.no

0

0
split.tensors.count

1224

1224

Tensor

Name

Type

Shape