GitHub
187K
Download
Models
Docs
Pricing
Sign in
Download
Models
Docs
Pricing
Sign in
eriktonon
/
step3.5-flash-uncensored
:196b
1
Download
Updated
5 hours ago
Cancel
196b
step3.5-flash-uncensored:196b
...
/
model
96e8f6af5df1 · 64GB
Metadata
general.architecture
step35
step35
general.file_type
IQ2_M
IQ2_M
step35.attention.head_count
[64, 96, 96, 96, 64, ...]
[64, 96, 96, 96, 64, ...]
step35.attention.head_count_kv
[8, 8, 8, 8, 8, ...]
[8, 8, 8, 8, 8, ...]
step35.attention.key_length
128
128
step35.attention.layer_norm_rms_epsilon
1e-05
1e-05
step35.attention.sliding_window
512
512
step35.attention.sliding_window_pattern
[false, true, true, true, false, ...]
[false, true, true, true, false, ...]
step35.attention.value_length
128
128
step35.block_count
45
45
step35.context_length
262144
262144
step35.embedding_length
4096
4096
step35.expert_count
288
288
step35.expert_feed_forward_length
1280
1280
step35.expert_gating_func
2
2
step35.expert_shared_feed_forward_length
1280
1280
step35.expert_used_count
8
8
step35.expert_weights_norm
true
true
step35.expert_weights_scale
3
3
step35.feed_forward_length
11264
11264
step35.leading_dense_block_count
3
3
step35.moe_every_n_layers
1
1
step35.rope.freq_base
5e+06
5e+06
step35.rope.freq_base_swa
10000
10000
step35.swiglu_clamp_exp
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
step35.swiglu_clamp_shexp
[0, 0, 0, 0, 0, ...]
[0, 0, 0, 0, 0, ...]
tokenizer.ggml.add_bos_token
true
true
tokenizer.ggml.add_sep_token
false
false
tokenizer.ggml.bos_token_id
0
0
tokenizer.ggml.eos_token_id
128007
128007
tokenizer.ggml.merges
[Ġ t, Ġ a, i n, Ġ Ġ, h e, ...]
[Ġ t, Ġ a, i n, Ġ Ġ, h e, ...]
tokenizer.ggml.model
gpt2
gpt2
tokenizer.ggml.padding_token_id
1
1
tokenizer.ggml.pre
deepseek-v3
deepseek-v3
tokenizer.ggml.token_type
[3, 3, 3, 1, 1, ...]
[3, 3, 3, 1, 1, ...]
tokenizer.ggml.tokens
[<|begin▁of▁sentence|>, <|end▁of▁sentence|>, <|▁pad▁|>, !, ", ...]
[<|begin▁of▁sentence|>, <|end▁of▁sentence|>, <|▁pad▁|>, !, ", ...]
mradermacher.convert_type
hf
hf
mradermacher.quantize_version
2
2
mradermacher.quantized_at
2026-02-22T14:09:25+01:00
2026-02-22T14:09:25+01:00
mradermacher.quantized_by
mradermacher
mradermacher
mradermacher.quantized_on
nico1
nico1
quantize.imatrix.chunks_count
315
315
quantize.imatrix.dataset
imatrix-training-full-3
imatrix-training-full-3
quantize.imatrix.entries_count
528
528
quantize.imatrix.file
Step-3.5-Flash-Ablitirated-i1-GGUF/Step-3.5-Flash-Ablitirated.imatrix.gguf
Step-3.5-Flash-Ablitirated-i1-GGUF/Step-3.5-Flash-Ablitirated.imatrix.gguf
Tensor
Name
Type
Shape
token_embd.weight
IQ3_S
IQ3_S
[4096, 128896]
blk.0
blk.0.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.0.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.0.attn_k_norm.weight
F32
F32
[128]
blk.0.attn_norm.weight
F32
F32
[4096]
blk.0.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.0.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.0.attn_q_norm.weight
F32
F32
[128]
blk.0.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.0.ffn_down.weight
IQ3_S
IQ3_S
[11264, 4096]
blk.0.ffn_gate.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.0.ffn_norm.weight
F32
F32
[4096]
blk.0.ffn_up.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.1
blk.1.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.1.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.1.attn_k_norm.weight
F32
F32
[128]
blk.1.attn_norm.weight
F32
F32
[4096]
blk.1.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.1.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.1.attn_q_norm.weight
F32
F32
[128]
blk.1.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.1.ffn_down.weight
IQ3_S
IQ3_S
[11264, 4096]
blk.1.ffn_gate.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.1.ffn_norm.weight
F32
F32
[4096]
blk.1.ffn_up.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.2
blk.2.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.2.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.2.attn_k_norm.weight
F32
F32
[128]
blk.2.attn_norm.weight
F32
F32
[4096]
blk.2.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.2.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.2.attn_q_norm.weight
F32
F32
[128]
blk.2.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.2.ffn_down.weight
IQ3_S
IQ3_S
[11264, 4096]
blk.2.ffn_gate.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.2.ffn_norm.weight
F32
F32
[4096]
blk.2.ffn_up.weight
IQ2_S
IQ2_S
[4096, 11264]
blk.3
blk.3.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.3.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.3.attn_k_norm.weight
F32
F32
[128]
blk.3.attn_norm.weight
F32
F32
[4096]
blk.3.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.3.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.3.attn_q_norm.weight
F32
F32
[128]
blk.3.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.3.exp_probs_b.bias
F32
F32
[288]
blk.3.ffn_down_exps.weight
IQ3_S
IQ3_S
[1280, 4096, 288]
blk.3.ffn_down_shexp.weight
IQ3_S
IQ3_S
[1280, 4096]
blk.3.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.3.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.3.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.3.ffn_norm.weight
F32
F32
[4096]
blk.3.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.3.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.4
blk.4.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.4.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.4.attn_k_norm.weight
F32
F32
[128]
blk.4.attn_norm.weight
F32
F32
[4096]
blk.4.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.4.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.4.attn_q_norm.weight
F32
F32
[128]
blk.4.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.4.exp_probs_b.bias
F32
F32
[288]
blk.4.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.4.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.4.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.4.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.4.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.4.ffn_norm.weight
F32
F32
[4096]
blk.4.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.4.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.5
blk.5.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.5.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.5.attn_k_norm.weight
F32
F32
[128]
blk.5.attn_norm.weight
F32
F32
[4096]
blk.5.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.5.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.5.attn_q_norm.weight
F32
F32
[128]
blk.5.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.5.exp_probs_b.bias
F32
F32
[288]
blk.5.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.5.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.5.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.5.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.5.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.5.ffn_norm.weight
F32
F32
[4096]
blk.5.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.5.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.6
blk.6.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.6.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.6.attn_k_norm.weight
F32
F32
[128]
blk.6.attn_norm.weight
F32
F32
[4096]
blk.6.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.6.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.6.attn_q_norm.weight
F32
F32
[128]
blk.6.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.6.exp_probs_b.bias
F32
F32
[288]
blk.6.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.6.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.6.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.6.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.6.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.6.ffn_norm.weight
F32
F32
[4096]
blk.6.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.6.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.7
blk.7.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.7.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.7.attn_k_norm.weight
F32
F32
[128]
blk.7.attn_norm.weight
F32
F32
[4096]
blk.7.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.7.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.7.attn_q_norm.weight
F32
F32
[128]
blk.7.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.7.exp_probs_b.bias
F32
F32
[288]
blk.7.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.7.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.7.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.7.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.7.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.7.ffn_norm.weight
F32
F32
[4096]
blk.7.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.7.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.8
blk.8.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.8.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.8.attn_k_norm.weight
F32
F32
[128]
blk.8.attn_norm.weight
F32
F32
[4096]
blk.8.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.8.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.8.attn_q_norm.weight
F32
F32
[128]
blk.8.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.8.exp_probs_b.bias
F32
F32
[288]
blk.8.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.8.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.8.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.8.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.8.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.8.ffn_norm.weight
F32
F32
[4096]
blk.8.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.8.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.9
blk.9.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.9.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.9.attn_k_norm.weight
F32
F32
[128]
blk.9.attn_norm.weight
F32
F32
[4096]
blk.9.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.9.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.9.attn_q_norm.weight
F32
F32
[128]
blk.9.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.9.exp_probs_b.bias
F32
F32
[288]
blk.9.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.9.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.9.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.9.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.9.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.9.ffn_norm.weight
F32
F32
[4096]
blk.9.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.9.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.10
blk.10.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.10.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.10.attn_k_norm.weight
F32
F32
[128]
blk.10.attn_norm.weight
F32
F32
[4096]
blk.10.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.10.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.10.attn_q_norm.weight
F32
F32
[128]
blk.10.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.10.exp_probs_b.bias
F32
F32
[288]
blk.10.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.10.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.10.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.10.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.10.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.10.ffn_norm.weight
F32
F32
[4096]
blk.10.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.10.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.11
blk.11.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.11.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.11.attn_k_norm.weight
F32
F32
[128]
blk.11.attn_norm.weight
F32
F32
[4096]
blk.11.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.11.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.11.attn_q_norm.weight
F32
F32
[128]
blk.11.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.11.exp_probs_b.bias
F32
F32
[288]
blk.11.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.11.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.11.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.11.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.11.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.11.ffn_norm.weight
F32
F32
[4096]
blk.11.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.11.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.12
blk.12.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.12.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.12.attn_k_norm.weight
F32
F32
[128]
blk.12.attn_norm.weight
F32
F32
[4096]
blk.12.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.12.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.12.attn_q_norm.weight
F32
F32
[128]
blk.12.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.12.exp_probs_b.bias
F32
F32
[288]
blk.12.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.12.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.12.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.12.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.12.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.12.ffn_norm.weight
F32
F32
[4096]
blk.12.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.12.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.13
blk.13.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.13.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.13.attn_k_norm.weight
F32
F32
[128]
blk.13.attn_norm.weight
F32
F32
[4096]
blk.13.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.13.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.13.attn_q_norm.weight
F32
F32
[128]
blk.13.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.13.exp_probs_b.bias
F32
F32
[288]
blk.13.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.13.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.13.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.13.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.13.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.13.ffn_norm.weight
F32
F32
[4096]
blk.13.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.13.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.14
blk.14.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.14.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.14.attn_k_norm.weight
F32
F32
[128]
blk.14.attn_norm.weight
F32
F32
[4096]
blk.14.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.14.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.14.attn_q_norm.weight
F32
F32
[128]
blk.14.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.14.exp_probs_b.bias
F32
F32
[288]
blk.14.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.14.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.14.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.14.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.14.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.14.ffn_norm.weight
F32
F32
[4096]
blk.14.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.14.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.15
blk.15.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.15.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.15.attn_k_norm.weight
F32
F32
[128]
blk.15.attn_norm.weight
F32
F32
[4096]
blk.15.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.15.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.15.attn_q_norm.weight
F32
F32
[128]
blk.15.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.15.exp_probs_b.bias
F32
F32
[288]
blk.15.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.15.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.15.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.15.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.15.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.15.ffn_norm.weight
F32
F32
[4096]
blk.15.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.15.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.16
blk.16.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.16.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.16.attn_k_norm.weight
F32
F32
[128]
blk.16.attn_norm.weight
F32
F32
[4096]
blk.16.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.16.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.16.attn_q_norm.weight
F32
F32
[128]
blk.16.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.16.exp_probs_b.bias
F32
F32
[288]
blk.16.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.16.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.16.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.16.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.16.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.16.ffn_norm.weight
F32
F32
[4096]
blk.16.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.16.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.17
blk.17.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.17.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.17.attn_k_norm.weight
F32
F32
[128]
blk.17.attn_norm.weight
F32
F32
[4096]
blk.17.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.17.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.17.attn_q_norm.weight
F32
F32
[128]
blk.17.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.17.exp_probs_b.bias
F32
F32
[288]
blk.17.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.17.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.17.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.17.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.17.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.17.ffn_norm.weight
F32
F32
[4096]
blk.17.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.17.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.18
blk.18.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.18.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.18.attn_k_norm.weight
F32
F32
[128]
blk.18.attn_norm.weight
F32
F32
[4096]
blk.18.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.18.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.18.attn_q_norm.weight
F32
F32
[128]
blk.18.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.18.exp_probs_b.bias
F32
F32
[288]
blk.18.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.18.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.18.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.18.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.18.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.18.ffn_norm.weight
F32
F32
[4096]
blk.18.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.18.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.19
blk.19.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.19.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.19.attn_k_norm.weight
F32
F32
[128]
blk.19.attn_norm.weight
F32
F32
[4096]
blk.19.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.19.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.19.attn_q_norm.weight
F32
F32
[128]
blk.19.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.19.exp_probs_b.bias
F32
F32
[288]
blk.19.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.19.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.19.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.19.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.19.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.19.ffn_norm.weight
F32
F32
[4096]
blk.19.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.19.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.20
blk.20.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.20.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.20.attn_k_norm.weight
F32
F32
[128]
blk.20.attn_norm.weight
F32
F32
[4096]
blk.20.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.20.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.20.attn_q_norm.weight
F32
F32
[128]
blk.20.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.20.exp_probs_b.bias
F32
F32
[288]
blk.20.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.20.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.20.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.20.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.20.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.20.ffn_norm.weight
F32
F32
[4096]
blk.20.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.20.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.21
blk.21.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.21.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.21.attn_k_norm.weight
F32
F32
[128]
blk.21.attn_norm.weight
F32
F32
[4096]
blk.21.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.21.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.21.attn_q_norm.weight
F32
F32
[128]
blk.21.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.21.exp_probs_b.bias
F32
F32
[288]
blk.21.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.21.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.21.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.21.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.21.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.21.ffn_norm.weight
F32
F32
[4096]
blk.21.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.21.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.22
blk.22.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.22.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.22.attn_k_norm.weight
F32
F32
[128]
blk.22.attn_norm.weight
F32
F32
[4096]
blk.22.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.22.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.22.attn_q_norm.weight
F32
F32
[128]
blk.22.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.22.exp_probs_b.bias
F32
F32
[288]
blk.22.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.22.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.22.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.22.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.22.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.22.ffn_norm.weight
F32
F32
[4096]
blk.22.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.22.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.23
blk.23.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.23.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.23.attn_k_norm.weight
F32
F32
[128]
blk.23.attn_norm.weight
F32
F32
[4096]
blk.23.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.23.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.23.attn_q_norm.weight
F32
F32
[128]
blk.23.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.23.exp_probs_b.bias
F32
F32
[288]
blk.23.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.23.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.23.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.23.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.23.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.23.ffn_norm.weight
F32
F32
[4096]
blk.23.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.23.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.24
blk.24.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.24.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.24.attn_k_norm.weight
F32
F32
[128]
blk.24.attn_norm.weight
F32
F32
[4096]
blk.24.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.24.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.24.attn_q_norm.weight
F32
F32
[128]
blk.24.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.24.exp_probs_b.bias
F32
F32
[288]
blk.24.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.24.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.24.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.24.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.24.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.24.ffn_norm.weight
F32
F32
[4096]
blk.24.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.24.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.25
blk.25.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.25.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.25.attn_k_norm.weight
F32
F32
[128]
blk.25.attn_norm.weight
F32
F32
[4096]
blk.25.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.25.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.25.attn_q_norm.weight
F32
F32
[128]
blk.25.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.25.exp_probs_b.bias
F32
F32
[288]
blk.25.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.25.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.25.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.25.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.25.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.25.ffn_norm.weight
F32
F32
[4096]
blk.25.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.25.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.26
blk.26.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.26.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.26.attn_k_norm.weight
F32
F32
[128]
blk.26.attn_norm.weight
F32
F32
[4096]
blk.26.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.26.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.26.attn_q_norm.weight
F32
F32
[128]
blk.26.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.26.exp_probs_b.bias
F32
F32
[288]
blk.26.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.26.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.26.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.26.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.26.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.26.ffn_norm.weight
F32
F32
[4096]
blk.26.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.26.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.27
blk.27.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.27.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.27.attn_k_norm.weight
F32
F32
[128]
blk.27.attn_norm.weight
F32
F32
[4096]
blk.27.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.27.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.27.attn_q_norm.weight
F32
F32
[128]
blk.27.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.27.exp_probs_b.bias
F32
F32
[288]
blk.27.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.27.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.27.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.27.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.27.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.27.ffn_norm.weight
F32
F32
[4096]
blk.27.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.27.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.28
blk.28.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.28.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.28.attn_k_norm.weight
F32
F32
[128]
blk.28.attn_norm.weight
F32
F32
[4096]
blk.28.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.28.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.28.attn_q_norm.weight
F32
F32
[128]
blk.28.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.28.exp_probs_b.bias
F32
F32
[288]
blk.28.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.28.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.28.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.28.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.28.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.28.ffn_norm.weight
F32
F32
[4096]
blk.28.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.28.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.29
blk.29.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.29.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.29.attn_k_norm.weight
F32
F32
[128]
blk.29.attn_norm.weight
F32
F32
[4096]
blk.29.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.29.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.29.attn_q_norm.weight
F32
F32
[128]
blk.29.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.29.exp_probs_b.bias
F32
F32
[288]
blk.29.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.29.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.29.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.29.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.29.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.29.ffn_norm.weight
F32
F32
[4096]
blk.29.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.29.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.30
blk.30.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.30.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.30.attn_k_norm.weight
F32
F32
[128]
blk.30.attn_norm.weight
F32
F32
[4096]
blk.30.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.30.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.30.attn_q_norm.weight
F32
F32
[128]
blk.30.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.30.exp_probs_b.bias
F32
F32
[288]
blk.30.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.30.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.30.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.30.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.30.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.30.ffn_norm.weight
F32
F32
[4096]
blk.30.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.30.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.31
blk.31.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.31.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.31.attn_k_norm.weight
F32
F32
[128]
blk.31.attn_norm.weight
F32
F32
[4096]
blk.31.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.31.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.31.attn_q_norm.weight
F32
F32
[128]
blk.31.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.31.exp_probs_b.bias
F32
F32
[288]
blk.31.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.31.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.31.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.31.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.31.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.31.ffn_norm.weight
F32
F32
[4096]
blk.31.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.31.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.32
blk.32.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.32.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.32.attn_k_norm.weight
F32
F32
[128]
blk.32.attn_norm.weight
F32
F32
[4096]
blk.32.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.32.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.32.attn_q_norm.weight
F32
F32
[128]
blk.32.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.32.exp_probs_b.bias
F32
F32
[288]
blk.32.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.32.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.32.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.32.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.32.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.32.ffn_norm.weight
F32
F32
[4096]
blk.32.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.32.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.33
blk.33.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.33.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.33.attn_k_norm.weight
F32
F32
[128]
blk.33.attn_norm.weight
F32
F32
[4096]
blk.33.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.33.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.33.attn_q_norm.weight
F32
F32
[128]
blk.33.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.33.exp_probs_b.bias
F32
F32
[288]
blk.33.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.33.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.33.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.33.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.33.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.33.ffn_norm.weight
F32
F32
[4096]
blk.33.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.33.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.34
blk.34.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.34.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.34.attn_k_norm.weight
F32
F32
[128]
blk.34.attn_norm.weight
F32
F32
[4096]
blk.34.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.34.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.34.attn_q_norm.weight
F32
F32
[128]
blk.34.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.34.exp_probs_b.bias
F32
F32
[288]
blk.34.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.34.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.34.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.34.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.34.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.34.ffn_norm.weight
F32
F32
[4096]
blk.34.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.34.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.35
blk.35.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.35.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.35.attn_k_norm.weight
F32
F32
[128]
blk.35.attn_norm.weight
F32
F32
[4096]
blk.35.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.35.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.35.attn_q_norm.weight
F32
F32
[128]
blk.35.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.35.exp_probs_b.bias
F32
F32
[288]
blk.35.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.35.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.35.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.35.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.35.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.35.ffn_norm.weight
F32
F32
[4096]
blk.35.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.35.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.36
blk.36.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.36.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.36.attn_k_norm.weight
F32
F32
[128]
blk.36.attn_norm.weight
F32
F32
[4096]
blk.36.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.36.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.36.attn_q_norm.weight
F32
F32
[128]
blk.36.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.36.exp_probs_b.bias
F32
F32
[288]
blk.36.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.36.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.36.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.36.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.36.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.36.ffn_norm.weight
F32
F32
[4096]
blk.36.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.36.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.37
blk.37.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.37.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.37.attn_k_norm.weight
F32
F32
[128]
blk.37.attn_norm.weight
F32
F32
[4096]
blk.37.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.37.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.37.attn_q_norm.weight
F32
F32
[128]
blk.37.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.37.exp_probs_b.bias
F32
F32
[288]
blk.37.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.37.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.37.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.37.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.37.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.37.ffn_norm.weight
F32
F32
[4096]
blk.37.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.37.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.38
blk.38.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.38.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.38.attn_k_norm.weight
F32
F32
[128]
blk.38.attn_norm.weight
F32
F32
[4096]
blk.38.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.38.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.38.attn_q_norm.weight
F32
F32
[128]
blk.38.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.38.exp_probs_b.bias
F32
F32
[288]
blk.38.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.38.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.38.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.38.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.38.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.38.ffn_norm.weight
F32
F32
[4096]
blk.38.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.38.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.39
blk.39.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.39.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.39.attn_k_norm.weight
F32
F32
[128]
blk.39.attn_norm.weight
F32
F32
[4096]
blk.39.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.39.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.39.attn_q_norm.weight
F32
F32
[128]
blk.39.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.39.exp_probs_b.bias
F32
F32
[288]
blk.39.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.39.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.39.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.39.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.39.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.39.ffn_norm.weight
F32
F32
[4096]
blk.39.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.39.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.40
blk.40.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.40.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.40.attn_k_norm.weight
F32
F32
[128]
blk.40.attn_norm.weight
F32
F32
[4096]
blk.40.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.40.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.40.attn_q_norm.weight
F32
F32
[128]
blk.40.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.40.exp_probs_b.bias
F32
F32
[288]
blk.40.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.40.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.40.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.40.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.40.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.40.ffn_norm.weight
F32
F32
[4096]
blk.40.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.40.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.41
blk.41.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.41.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.41.attn_k_norm.weight
F32
F32
[128]
blk.41.attn_norm.weight
F32
F32
[4096]
blk.41.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.41.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.41.attn_q_norm.weight
F32
F32
[128]
blk.41.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.41.exp_probs_b.bias
F32
F32
[288]
blk.41.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.41.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.41.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.41.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.41.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.41.ffn_norm.weight
F32
F32
[4096]
blk.41.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.41.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.42
blk.42.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.42.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.42.attn_k_norm.weight
F32
F32
[128]
blk.42.attn_norm.weight
F32
F32
[4096]
blk.42.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.42.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.42.attn_q_norm.weight
F32
F32
[128]
blk.42.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.42.exp_probs_b.bias
F32
F32
[288]
blk.42.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.42.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.42.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.42.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.42.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.42.ffn_norm.weight
F32
F32
[4096]
blk.42.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.42.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.43
blk.43.attn_gate.weight
IQ2_S
IQ2_S
[4096, 96]
blk.43.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.43.attn_k_norm.weight
F32
F32
[128]
blk.43.attn_norm.weight
F32
F32
[4096]
blk.43.attn_output.weight
IQ3_S
IQ3_S
[12288, 4096]
blk.43.attn_q.weight
IQ2_S
IQ2_S
[4096, 12288]
blk.43.attn_q_norm.weight
F32
F32
[128]
blk.43.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.43.exp_probs_b.bias
F32
F32
[288]
blk.43.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.43.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.43.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.43.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.43.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.43.ffn_norm.weight
F32
F32
[4096]
blk.43.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.43.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.44
blk.44.attn_gate.weight
IQ2_S
IQ2_S
[4096, 64]
blk.44.attn_k.weight
IQ2_S
IQ2_S
[4096, 1024]
blk.44.attn_k_norm.weight
F32
F32
[128]
blk.44.attn_norm.weight
F32
F32
[4096]
blk.44.attn_output.weight
IQ3_S
IQ3_S
[8192, 4096]
blk.44.attn_q.weight
IQ2_S
IQ2_S
[4096, 8192]
blk.44.attn_q_norm.weight
F32
F32
[128]
blk.44.attn_v.weight
Q4_K
Q4_K
[4096, 1024]
blk.44.exp_probs_b.bias
F32
F32
[288]
blk.44.ffn_down_exps.weight
IQ2_S
IQ2_S
[1280, 4096, 288]
blk.44.ffn_down_shexp.weight
IQ2_S
IQ2_S
[1280, 4096]
blk.44.ffn_gate_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.44.ffn_gate_inp.weight
F32
F32
[4096, 288]
blk.44.ffn_gate_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
blk.44.ffn_norm.weight
F32
F32
[4096]
blk.44.ffn_up_exps.weight
IQ2_S
IQ2_S
[4096, 1280, 288]
blk.44.ffn_up_shexp.weight
IQ2_S
IQ2_S
[4096, 1280]
output.weight
Q5_K
Q5_K
[4096, 128896]
rope_freqs.weight
F32
F32
[64]
output_norm.weight
F32
F32
[4096]