feat: support bloom-1b4-zh

examples/CMakeLists.txt

@@ -28,3 +28,4 @@ add_subdirectory(replit)
 add_subdirectory(mpt)
 add_subdirectory(starcoder)
 add_subdirectory(sam)
+add_subdirectory(bloom-1b4-zh)

examples/bloom-1b4-zh/CMakeLists.txt

@@ -0,0 +1,13 @@
+#
+# bloom-1b4-zh
+
+set(TEST_TARGET bloom-1b4-zh)
+add_executable(${TEST_TARGET} main.cpp)
+target_link_libraries(${TEST_TARGET} PRIVATE ggml common common-ggml)
+
+#
+# bloom-1b4-zh-quantize
+
+set(TEST_TARGET bloom-1b4-zh-quantize)
+add_executable(${TEST_TARGET} quantize.cpp)
+target_link_libraries(${TEST_TARGET} PRIVATE ggml common common-ggml)

examples/bloom-1b4-zh/README.md

@@ -0,0 +1,27 @@
+# Bloom-1b4-zh
+
+Ref: https://huggingface.co/Langboat/bloom-1b4-zh
+
+## Usage
+
+```bash
+# get the repo and build it
+git clone https://github.com/ggerganov/ggml
+cd ggml
+mkdir build && cd build
+cmake ..
+make -j
+
+# get the model from HuggingFace
+# be sure to have git-lfs installed
+git clone https://huggingface.co/Langboat/bloom-1b4-zh
+
+# convert model to FP16
+mkdir -p output
+python3 ../examples/bloom-1b4-zh/convert-h5-to-ggml.py ./bloom-1b4-zh ./output
+
+# run inference using FP16 precision
+./bin/bloom-1b4-zh -m ./output/ggml-model-f16.bin -p "I believe the meaning of life is" -t 8 -n 64
+
+# quantize the model to 5-bits using Q5_0 quantization
+./bin/bloom-1b4-zh-quantize ./output/ggml-model-f16.bin ./output/ggml-model-q5_0.bin q5_0

examples/bloom-1b4-zh/convert-h5-to-ggml.py

@@ -0,0 +1,211 @@
+# Convert bloom-1b4-zh h5 transformer model to ggml format
+#
+# Load the model using BloomForCausalLM.
+# Iterate over all variables and write them to a binary file.
+#
+# For each variable, write the following:
+#   - Number of dimensions (int)
+#   - Name length (int)
+#   - Dimensions (int[n_dims])
+#   - Name (char[name_length])
+#   - Data (float[n_dims])
+#
+# By default, the bigger matrices are converted to 16-bit floats.
+# This can be disabled by adding the "use-f32" CLI argument.
+#
+# At the start of the ggml file we write the model parameters
+# and vocabulary.
+#
+
+import sys
+import struct
+import json
+import numpy as np
+import re
+
+from transformers import BloomForCausalLM
+
+
+# ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
+def bytes_to_unicode():
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a signficant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1))
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8+n)
+            n += 1
+    cs = [chr(n) for n in cs]
+    return dict(zip(bs, cs))
+
+
+# Download model from https://huggingface.co/Langboat/bloom-1b4-zh
+if len(sys.argv) < 3:
+    print("Usage: convert-h5-to-ggml.py dir-model dir-output [use-f32]\n")
+    sys.exit(1)
+
+
+# output in the same directory as the model
+dir_model = sys.argv[1]
+fname_out = sys.argv[2] + "/ggml-model-f16.bin"
+
+with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
+    vocab = json.load(f)
+    vocab = vocab['model']['vocab']
+
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+# use 16-bit or 32-bit floats
+use_f16 = True
+if len(sys.argv) > 3:
+    use_f16 = False
+    fname_out = sys.argv[2] + "/ggml-model-f32.bin"
+
+model = BloomForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True)
+
+list_vars = model.state_dict()
+# bloom-1b4-zh models share the WTE tensor as the LM head
+assert np.allclose(list_vars["transformer.word_embeddings.weight"].numpy(),
+                   list_vars["lm_head.weight"].numpy())
+del list_vars["lm_head.weight"]
+
+fout = open(fname_out, "wb")
+
+fout.write(struct.pack("i", 0x67676d6c))  # magic: ggml in hex
+fout.write(struct.pack("i", hparams["vocab_size"]))
+fout.write(struct.pack("i", hparams["seq_length"]))
+fout.write(struct.pack("i", hparams["hidden_size"]))
+fout.write(struct.pack("i", hparams["n_head"]))
+fout.write(struct.pack("i", hparams["n_layer"]))
+fout.write(struct.pack("i", use_f16))
+
+byte_encoder = bytes_to_unicode()
+byte_decoder = {v: k for k, v in byte_encoder.items()}
+
+fout.write(struct.pack("i", len(vocab)))
+
+for key in vocab:
+    text = bytearray([byte_decoder[c] for c in key])
+    fout.write(struct.pack("i", len(text)))
+    fout.write(text)
+
+for name in list_vars.keys():
+    data = list_vars[name].squeeze().numpy()
+    print("Processing variable: " + name + " with shape: ", data.shape)
+
+    n_dims = len(data.shape)
+
+    # ftype == 0 -> float32, ftype == 1 -> float16
+    ftype = 0
+    if use_f16:
+        if name[-7:] == ".weight" and n_dims == 2:
+            print("  Converting to float16")
+            data = data.astype(np.float16)
+            ftype = 1
+        else:
+            print("  Converting to float32")
+            data = data.astype(np.float32)
+            ftype = 0
+
+    # "transformer.h.*.mlp.dense_4h_to_h.weight" ==> "h.*.mlp.dense_4h_to_h.weight"  # noqa
+    if name.startswith("transformer."):
+        name = name[12:]
+
+    # rename headers to keep compatibility
+    if name == "word_embeddings_layernorm.weight":
+        name = "model/ln_wte/g"
+    elif name == "word_embeddings_layernorm.bias":
+        name = "model/ln_wte/b"
+    elif name == "word_embeddings.weight":
+        name = "model/wte"
+    elif re.match(r"h\.\d+\.input_layernorm\.weight", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/ln_1/g"
+    elif re.match(r"h\.\d+\.input_layernorm\.bias", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/ln_1/b"
+    elif re.match(r"h\.\d+\.self_attention\.query_key_value\.weight", name):
+        # Map bloom-style qkv_linear to gpt-style qkv_linear
+        # bloom: https://github.com/huggingface/transformers/blob/main/src/transformers/models/bloom/modeling_bloom.py#L238-L252  # noqa
+        # gpt-2: https://github.com/huggingface/transformers/blob/main/src/transformers/models/gpt2/modeling_gpt2.py#L312  # noqa
+        qkv_weights = data.reshape(
+            (hparams["n_head"], 3, hparams["hidden_size"] // hparams["n_head"],
+             hparams["hidden_size"])
+        )
+        data = np.concatenate(
+            (qkv_weights[:, 0, :, :].reshape((-1, hparams["hidden_size"])),
+             qkv_weights[:, 1, :, :].reshape((-1, hparams["hidden_size"])),
+             qkv_weights[:, 2, :, :].reshape((-1, hparams["hidden_size"]))),
+            axis=0
+        )
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/attn/c_attn/w"
+    elif re.match(r"h\.\d+\.self_attention\.query_key_value\.bias", name):
+        qkv_bias = data.reshape(
+            (hparams["n_head"], 3,
+             hparams["hidden_size"] // hparams["n_head"])
+        )
+        data = np.concatenate(
+            (qkv_bias[:, 0, :].reshape((hparams["hidden_size"],)),
+             qkv_bias[:, 1, :].reshape((hparams["hidden_size"],)),
+             qkv_bias[:, 2, :].reshape((hparams["hidden_size"],))),
+            axis=0
+        )
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/attn/c_attn/b"
+    elif re.match(r"h\.\d+\.self_attention\.dense\.weight", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/attn/c_proj/w"
+    elif re.match(r"h.\d+\.self_attention\.dense\.bias", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/attn/c_proj/b"
+    elif re.match(r"h.\d+\.post_attention_layernorm\.weight", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/ln_2/g"
+    elif re.match(r"h.\d+\.post_attention_layernorm\.bias", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/ln_2/b"
+    elif re.match(r"h.\d+\.mlp\.dense_h_to_4h\.weight", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/mlp/c_fc/w"
+    elif re.match(r"h.\d+\.mlp\.dense_h_to_4h\.bias", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/mlp/c_fc/b"
+    elif re.match(r"h.\d+\.mlp\.dense_4h_to_h\.weight", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/mlp/c_proj/w"
+    elif re.match(r"h.\d+\.mlp\.dense_4h_to_h\.bias", name):
+        i = re.findall("\d+", name)[0]
+        name = f"model/h{i}/mlp/c_proj/b"
+    elif name == "ln_f.weight":
+        name = "model/ln_f/g"
+    elif name == "ln_f.bias":
+        name = "model/ln_f/b"
+    else:
+        print("Unrecognized variable name. %s", name)
+
+    # header
+    str = name.encode('utf-8')
+    fout.write(struct.pack("iii", n_dims, len(str), ftype))
+    for i in range(n_dims):
+        fout.write(struct.pack("i", data.shape[n_dims - 1 - i]))
+    fout.write(str)
+
+    # data
+    data.tofile(fout)
+
+fout.close()
+
+print("Done. Output file: " + fname_out)
+print("")